)]}'
{
  "log": [
    {
      "commit": "33cfc031fd1b839e61fd333451ddc2cf855809aa",
      "tree": "c6c291e1dd5c31f3475456201db0347bf54f6a01",
      "parents": [
        "305d232f7bfdcdfcafdd7cd4f90bbfab6a2fa7a9",
        "e47542659dc1ab8bae12fcab7172b9c5751b4589"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Oct 28 04:42:36 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Oct 28 04:42:36 2026 -0700"
      },
      "message": "Merge \"Exclude org.json pulled in by Closure Templates\""
    },
    {
      "commit": "e47542659dc1ab8bae12fcab7172b9c5751b4589",
      "tree": "ca13deadb16e2302bd3df34cd283ddce11dee6bb",
      "parents": [
        "bcb875966dd0cceff9fd1e5c09b268f5ba49d19f"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Oct 24 16:47:52 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Wed Oct 28 02:23:22 2026 -0700"
      },
      "message": "Exclude org.json pulled in by Closure Templates\n\nThe gitiles SBOM scan (//:gitiles_sbom_scan, Trivy over the generated\nCycloneDX SBOM) flagged org.json:json 20230618 shipping in gitiles.war\nwith CVE-2023-5072 (HIGH): a parser confusion that can exhaust memory.\nIt is pulled in transitively by Closure Templates\n(com.google.template:soy).\n\nGitiles does not use org.json in its own code, and Gerrit excludes it\nfrom the same soy version and ships without it. So omit it here the same\nway we dropped end-of-life log4j 1.x: exclude the transitive jar from\nsoy and re-pin. The war no longer ships json-20230618.jar and the scan\nis clean.\n\nChange-Id: I2d8cc40b8c38fcf2e0f8392adf4e93d92ae2c404\n"
    },
    {
      "commit": "305d232f7bfdcdfcafdd7cd4f90bbfab6a2fa7a9",
      "tree": "dc93af7aa6711a8e8458ad74d3cfed62cadd246a",
      "parents": [
        "bc862b8d1975826c92e15dbdc1c5dd2eb4a90372",
        "bcb875966dd0cceff9fd1e5c09b268f5ba49d19f"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Oct 28 02:07:35 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Oct 28 02:07:35 2026 -0700"
      },
      "message": "Merge \"Exclude end-of-life log4j 1.x from the flogger backend\""
    },
    {
      "commit": "bc862b8d1975826c92e15dbdc1c5dd2eb4a90372",
      "tree": "d0178bffb81a838ac7058b7e1d4485671044c3b5",
      "parents": [
        "96c77914baf43cc274b69f2415a84529c6882974",
        "d26eac6e8d283656be680973098ac3d65768ba2c"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Mon Oct 26 17:15:10 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Mon Oct 26 17:15:10 2026 -0700"
      },
      "message": "Merge changes I82ec9852,I32240619\n\n* changes:\n  Search repositories with / on the host index\n  Add a keyboard-driven file finder to the Gitiles header\n"
    },
    {
      "commit": "bcb875966dd0cceff9fd1e5c09b268f5ba49d19f",
      "tree": "ffd69e507bdbfa3db9c91936cb5a5f73ca4b8595",
      "parents": [
        "96c77914baf43cc274b69f2415a84529c6882974"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Oct 24 16:35:30 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Oct 24 16:56:01 2026 +0200"
      },
      "message": "Exclude end-of-life log4j 1.x from the flogger backend\n\nAdding the CycloneDX SBOM scan for gitiles (//:gitiles_sbom_scan, Trivy\nover the generated SBOM) surfaced log4j:log4j 1.2.15 shipping in\ngitiles.war with six CVEs, three of them CRITICAL: CVE-2019-17571,\nCVE-2022-23305 and CVE-2022-23307, plus CVE-2021-4104, CVE-2022-23302 and\nCVE-2023-26464. This is end-of-life log4j 1.x with no fixed release.\n\nflogger-log4j-backend pulls log4j:log4j in transitively. reload4j, the\npatched drop-in fork, is already a direct dependency and provides the same\nclasses, so exclude log4j:log4j from the backend\u0027s transitive closure and\nre-pin. gitiles.war now ships reload4j-1.2.26.jar alone; the vulnerable\nlog4j-1.2.15.jar is gone. Gerrit keeps log4j 1.x out the same way.\n\nChange-Id: I09b20eb11cae70acec87cccf96b82acf1f0eb3c6\n"
    },
    {
      "commit": "d26eac6e8d283656be680973098ac3d65768ba2c",
      "tree": "50fcb3c71347e1936bb91ff3bc9d64dc3b3bd4e4",
      "parents": [
        "34ca370f8d0f897cd7f307c7bedfb1aa8ff7c5a8"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 22 19:09:10 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Oct 23 00:31:57 2026 +0000"
      },
      "message": "Search repositories with / on the host index\n\nThe finder answers \"/\" on every page that addresses a tree, but not on\nthe host index -- which is where readers arrive, and which is itself a\nlisting, of repositories rather than files. A keystroke that works\nnearly everywhere and then silently does nothing never becomes muscle\nmemory.\n\nRender the same palette there, over repository names. The page already\ncontains every repository it lists, so the finder reads them out of the\ndocument instead of asking for the same names again: a host index\nresponse names no revision and is therefore served no-store, so a\nsecond request would be made on every page load and cached after none\nof them. Reading the page also makes the finder\u0027s answers agree with it\nby construction -- nothing can be missing from the finder that the\nreader can already see.\n\nThe worker gains a load variant taking a list rather than a URL, and\nsorts it. That is not cosmetic: two of its lookups are binary searches\nover a byte-ordered index. indexOfPath is one, so an unsorted list\nsilently fails to find names that are present -- 2,828 of 2,838 on a\nreal host index. The literal-prefix channel is the other, so an\nunsorted list would leave a typed namespace unpromoted, which on a host\nindex is the common case: readers type \"platform/frameworks\", not a\nrepository\u0027s last segment. The directory-run compression additionally\nassumes sorted input for its speed. A git tree listing arrives sorted;\na list scraped from a page cannot be assumed to be.\n\nRepository results are never warmed. Warming is only worth doing\nbecause file links are SHA-pinned and therefore cacheable; a repository\nindex page names no revision, so a speculative fetch of one would be\ndiscarded and the request wasted.\n\nDescriptions are shown beside the name, as the page shows them, because\nrepository names alone are often not enough to choose between.\n\nVerified against a host index of 2,838 real repository names. Engine:\nevery name is found by typing it in full, none is absent from a search\nfor its own last segment, all 2,838 are located by indexOfPath, a\nshuffled list yields an identical index, and skipping the sort\ndemonstrably breaks it. Browser: the index is built with zero network\nrequests, rows link where the page links, nothing is warmed, Enter\nopens the repository, and returning offers it again. Keystroke to\nrepaint on that index is 3.9 ms median, 6.0 ms worst.\n\nRepository search sits inside the gitiles.fileSearch guard, so the one\nswitch covers both listings; FileSearchTest and Documentation/config.md\nreflect that.\n\nThree unit tests cover the new load path: that a supplied listing is\nused without a fetch, that it is sorted whatever order it arrives in,\nand that the sort is by UTF-8 bytes rather than UTF-16 code units. The\nlast one matters above the basic plane, where the two orders disagree\n-- Array#sort would put U+10000 before U+FFFD, the bytes run the other\nway -- and it discriminates: replacing the sort entirely fails two\ntests, replacing it with Array#sort fails exactly that one.\n\nEight more cover repository mode in the palette: that the listing is\nread out of the page and nothing is fetched, that rows link where the\npage links and carry the description, that repositories named after\nObject.prototype properties survive with or without a description, that\nEnter navigates and records the visit under the host index key, that an\nindex reached with or without a trailing slash keeps one history, that\nno repository page is ever warmed, that an empty index hides the trigger\nand offers no palette, and that the same repository listed twice is\noffered once. Reverted fixes were each caught -- building the name or\ndescription maps with an object literal instead of Object.create(null),\ndropping the repository-mode return from schedulePrefetch, dropping the\nduplicate skip, and keying history on the raw pathname. Unmutated, the\npalette suite passes 43 of 43 and the worker suite 21 of 21.\n\nBug: 564986686\nTAG\u003dagy\nCONV\u003d0b3e4b30-d902-4587-8bb2-ce23109d6955\nCONV\u003d60d1bfdf-34a4-460e-85ab-0f251e052543\nChange-Id: I82ec98521c722a6f316eed456e1894e5b7da98c7\n"
    },
    {
      "commit": "34ca370f8d0f897cd7f307c7bedfb1aa8ff7c5a8",
      "tree": "f75210bc51b1240ddac31a115a5c6b0c1aa99368",
      "parents": [
        "e0d8b45eb4afd832309243f743fd59925d925750"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 22 19:09:10 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Thu Oct 22 17:34:05 2026 +0000"
      },
      "message": "Add a keyboard-driven file finder to the Gitiles header\n\nPressing \"/\" on any page that addresses a commit opens a palette that\nfuzzy-matches every blob path in that revision; Enter navigates\nstraight to the file.\n\nThe index is the paths_only tree listing, fetched on first open rather\nthan on page load, so a reader who never searches costs the server\nnothing. Its URL is pinned to the resolved commit SHA: BaseServlet\nonly marks a response cacheable when the revision is named by object\nID, so a branch-named URL would be refetched on every open, and\ncontent addressing retires a stale copy without a revalidation\nroundtrip.\n\nMatching runs in a Worker against a string-free index -- path bytes in\nflat buffers with Int32Array offsets -- so no amount of index work can\ndrop a frame on the main thread. Directories are interned, collapsing\nchromium/src from 506,237 paths to 46,286 directory runs, and the byte\nbuffers are trimmed to what was written. The whole index is then 28 MB\nagainst 37 MB of path text: smaller than the corpus it indexes.\n\nThere is no cap. Every path in the revision is held and considered; if\na path matches, either it is on screen or twenty strictly\nbetter-ranked paths are.\n\nRanking\n-------\n\nQueries run in tiers, cheapest first, and a tier is skipped only when\na cheaper one has already filled the display. That is exact rather\nthan heuristic, because the tiers are a strict ordering: nothing a\nlater one produces can displace what an earlier one found.\n\n  A  the whole path literally starts with the query\n  B  the basename starts with it\n  C  the basename contains it as a subsequence\n  D  the path contains it as a subsequence\n\nA comes first because B, C and D all score the basename alone, and a\nquery the directory has already absorbed leaves the basename scoring\nnothing. Without A, typing a path got monotonically worse the more of\nit you spelled: 1,254 paths start with \"base/t\" and none of them\nreached the top twenty. A is two binary searches, because the index is\nin byte order and so those paths are contiguous. It is case-sensitive\nby design -- a case-folded range would not be contiguous -- which is\nsound only because it promotes and never filters, so a reader who\ntypes the wrong case loses the promotion, not the result.\n\nGreedy directory consumption decides whether a file matches, not how\nwell. Consuming as much of the query as possible in the directory\nleaves the shortest remainder for the basename, which is what makes D\nsound and cheap, but it is forced rather than chosen: the directory of\nbase/base64.h eats \"base\" out of the query \"base64.h\", so the file\nscored as a directory-assisted match and took D\u0027s penalty for that.\nTyping a filename exactly then put it at position 9, behind eight\nunrelated test pages whose directories happened to contain none of the\nquery. A basename that carries the whole query on its own is scored as\nthe basename match it is, whatever the directory could also have\nsupplied. Only the order changes; the candidate set, and the monotone\nnarrowing it rests on, are untouched.\n\nSubsequence matching is monotone under query extension, so each\nkeystroke narrows the previous candidate set and backspace pops it.\nDirectory prefixes are memoised, which is what makes D affordable: the\n506k paths in chromium/src share only 46k distinct directories. D\nresumes each directory\u0027s greedy match from the length it shares with\nthe previous one, rejects a file whose basename alone cannot supply\nthe characters the directory failed to consume, and accepts a whole\ndirectory run wholesale once the directory has consumed the query.\n\nMeasured on that corpus (506,237 blobs) under node v22, on the worker\nthread: 480 ms to build the index, and 25 ms for the worst keystroke,\nwhich is a basename query. Path-directed queries are answered by A in\nunder a millisecond. End to end in a browser on the same corpus,\nincluding fetch, gzip, JSON.parse and the postMessage round trip:\n1,737 ms from \"/\" to a usable index and a median keystroke of 1.9 ms.\nAndroid\u0027s largest repository, toolchain/rustc at 223,429 paths, is\n718 ms and 2.1 ms. Typing does not depend on the link: shaping the\norigin to 5 Mbps and 150 ms of RTT leaves the median at 2.0 ms and\nmoves only the one-time open, to 2.95 s.\n\nWhen A or B fills the display the engine stops there, so its count is\nthen a lower bound and the display says so with a trailing \"+\".\n\nMatched characters are marked in the row. The engine measures in UTF-8\nbytes and the UI marks substrings of a JS string, whose indices are\nUTF-16 code units, so the offsets are only meaningful for an ASCII\npath; on anything else they drifted and marked characters that had not\nmatched. A path whose byte length differs from its string length is\ntherefore shown unmarked. That is four paths in chromium/src, and the\nlink is built from the whole path and was never affected.\n\nEmpty state\n-----------\n\nAn empty box offers the files this reader keeps returning to, best\nfirst, rather than the repository in tree order -- which on\nchromium/src means offering \".gn\". The score is an exponentially\ndecayed visit count: on each visit\n\n    s \u003c- s * 2^(-dt / HALF_LIFE) + 1\n\nwhich expands to the sum over past visits of 2^(-age/HALF_LIFE), with\na one-week half-life, so frequent files outrank once-opened ones and\nabandoned files fall away. Remembered paths are checked against the\nindex before they are shown, so a file deleted since it was last\nopened is dropped rather than offered as a row that 404s. Only files\nopened through the finder are recorded. Ranking for non-empty queries\nis untouched, so two readers typing the same query see the same\nresults in the same order. The store lives in localStorage, is keyed\nper repository, holds at most 200 entries, and is a record of what\nthis reader opened, so on a shared machine it is visible to anyone\nwith the browser profile. Storage being absent, denied, full or\ncorrupt falls back to the plain listing. The offered list is cut to\nthe display limit like every other channel: the UI paints exactly that\nmany rows but takes its selection bounds from what it was sent, so a\nlonger list let the selection run off the end of the rows onto nothing.\n\nWarming\n-------\n\nThe finder resolves a query in single-digit milliseconds and then\nhands the reader a link they still have to wait on, so the highlighted\ndestination is warmed. This is only possible because result links are\nSHA-pinned and therefore cacheable. Three mechanisms were measured in\nChrome under the access pattern the finder creates -- many warms on\none page, no reload in between:\n\n  link rel\u003dprefetch  warms some destinations and cancels others with\n                     ERR_ABORTED; unusable here\n  fetch, low prio    completes every time, lands on disk, but carries\n                     no Sec-Purpose\n  speculation rules  completes every time, lands in the prefetch\n                     cache, carries Sec-Purpose: prefetch\n\nSo the warm is a one-URL prefetch rule where the API exists, falling\nback to a low-priority fetch where it does not. The rule set is\nswapped rather than grown, so at most one destination is speculated on\nat a time, and it is torn down when the palette closes. A 250 ms\ndebounce, a dedupe set and a cap of 32 distinct destinations per\nsession keep it polite.\n\nGitiles sets no Content-Security-Policy of its own, so the policy\nbelongs to whoever deploys it. A script-src carrying neither\n\u0027strict-dynamic\u0027 nor \u0027inline-speculation-rules\u0027 blocks the rule set,\nand a blocked rule reports nothing to the code that installed it, so\nwarming would stop with no signal. A securitypolicyviolation listener\ndowngrades to the fetch path, which connect-src governs and which\ntherefore survives exactly the policy that kills the rule.\n\nReach\n-----\n\nThe finder is on every page that names a repository, not only trees.\nA keyboard shortcut only earns its keep once it is reflex, and a\nreader who hits \"/\" on the repository index and gets nothing has\nlearned not to try. Pages that address a revision search that\nrevision; the repository index and the ref list carry none, so they\nfall back to HEAD, which on the index is the tree already on screen. A\nrevision that peels to something other than a commit deliberately does\nnot fall back rather than silently answer about different content.\nBecause the URL is content-addressed, every page in a repository names\nthe same listing, so it is fetched once and reused across navigations.\n\nArrow keys, Ctrl-N/P and PageUp/PageDown move the selection; Escape\ncloses; Enter opens the file and Ctrl/Cmd-Enter opens it in a new tab.\nThe palette is closed on pagehide so that a page restored from the\nback/forward cache does not come back with a stale overlay covering\nit.\n\nFocus never leaves the input, so the input names the selected row with\naria-activedescendant -- without it a screen reader is told nothing at\nall as the selection travels, which for a feature that exists to be\ndriven from the keyboard is most of the feature. Rows are not tab\nstops and Tab is held, so focus cannot walk out of a dialog that\ndeclares itself modal.\n\nConfiguration\n-------------\n\nSetting gitiles.fileSearch (default true) to false turns the finder\noff. The index is linear in the number of paths in the revision and\ndeliberately uncapped, because a file that exists must be findable, so\na deployment whose repositories are larger than anything measured here\nneeds a switch rather than a patch.\n\nA deployment can also decline it without meaning to: a policy with a\nrestrictive worker-src rejects the Worker constructor synchronously,\nwith no event to observe. That is caught, because the overlay is\nalready on screen and the input is not yet focused when it throws --\nunhandled, it left a dialog the reader could neither type into nor\nclose. The reader is told instead, and asking again retries, so a\none-off failure loading the index does not disable the finder for the\nlife of the page.\n\nVerification\n------------\n\nFileSearchTest pins the things that break silently: the listing URL\nnaming a commit SHA rather than a branch or tag object, the script tag\ncarrying the request CSP nonce, Markdown documentation pages rendering\nthe trigger and overlay, and gitiles.fileSearch \u003d false rendering no\nfinder at all.\n\nThe CSP downgrade was driven in headless Chrome against a served\nscript-src \u0027self\u0027, paired against the same proxy without the header:\nthe hardened run reports a script-src-elem violation, drops the rule\nand still warms the highlighted destination exactly once; the control\nreports no violation, keeps the rule and also warms exactly once.\n\nThe engine was checked against a brute-force oracle over all 506,237\npaths of chromium/src: every path is findable by its own basename, no\ndisplayed row fails to match, and a reported total is exact whenever\nthe engine claims exactness and a true lower bound otherwise. The\nresumed directory pass was additionally compared against an\nindependent per-directory rescan for 3,021 queries across all 46,286\ndirectories, 139,830,006 comparisons, with no disagreement. The\npalette, the empty state and the warming were driven end to end in\nheadless Chrome against the same repository.\n\nRanking, the empty state at more than a screenful, the unmarked\nnon-ASCII row, aria-activedescendant tracking the selection, Tab being\nheld, and a refused Worker constructor were each driven in headless\nChrome against chromium/src. Scoring a basename-only match as one\ncosts nothing measurable: over eight queries spanning every channel,\nthe median keystroke moves by at most 0.05 ms either way and the index\nbuild is unchanged.\n\nEighteen unit tests drive the worker through its message protocol on\nthe js_test lane, covering the listing load and its failure, ranking,\nthe promoted path-prefix and basename-prefix channels, negative-scored\nweak matches, directory highlight ranges, the unmarked non-ASCII row,\nleading U+FEFF preservation, and the cut to the displayed row count.\nEach was checked against a mutant: reverting the basename-first\nscoring fix, the ASCII guard on highlight ranges, or the cut to LIMIT\nmakes the suite fail, and the unmutated worker passes 18 of 18.\n\nThirty-five more drive the palette itself. jstests/fake_dom.js is not\na browser and does not try to be one: it is a recording of the calls\nthe script makes -- attributes set, classes toggled, preventDefault\ncounts, timers queued, storage written -- and what it declines to\nmodel is listed in its header. That is the right shape here because\nwhat is worth pinning is this script\u0027s own bookkeeping rather than any\nbrowser semantics. The tests cover opening and dismissal, row reuse\nand the cut to twenty, selection clamping and aria-activedescendant,\nIME composition guards, the worker handshake and the ways it can fail,\nthe frecency store\u0027s decay, cap and empty fallback, and warming both\nunder speculation rules and after a policy has refused them.\n\nNine reverted fixes were each caught: not resetting the selection on a\nnew query, clamping it to the result count rather than to the rows on\nscreen, leaving rows in the tab order, dropping the row ids, dropping\naria-activedescendant, letting Tab escape the palette, leaving the\nWorker constructor unguarded, dropping the exponential decay, and\nleaving the store uncapped. Unmutated, 35 of 35 pass.\n\nBug: 564986686\nTAG\u003dagy\nCONV\u003d0b3e4b30-d902-4587-8bb2-ce23109d6955\nCONV\u003d60d1bfdf-34a4-460e-85ab-0f251e052543\nChange-Id: I32240619cb19aedb30cbf82f96c926031a242d74\n"
    },
    {
      "commit": "96c77914baf43cc274b69f2415a84529c6882974",
      "tree": "86ac1e645914c22c144b5f87619615433d783ead",
      "parents": [
        "abe95e599a96a8e56fa427569c1fa88a68a51717",
        "8bd0a16180497358515ea28e92f8335d98300739"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Oct 21 19:34:03 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Oct 21 19:34:03 2026 -0700"
      },
      "message": "Merge \"Bazel: Align dependency versions with Gerrit\""
    },
    {
      "commit": "abe95e599a96a8e56fa427569c1fa88a68a51717",
      "tree": "7a24cd87e9d002ad82639c0ce2d4afe81403b421",
      "parents": [
        "1814dd26080128b3a7083afe6f8b2c141a4d282b",
        "c6aa350acb774c7880683926f3cd4a41a673c9ff"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Oct 21 19:33:37 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Oct 21 19:33:37 2026 -0700"
      },
      "message": "Merge \"Bump rules_jvm_external to 7.2 and drop root pinning workaround\""
    },
    {
      "commit": "1814dd26080128b3a7083afe6f8b2c141a4d282b",
      "tree": "e83bc8c89311756aebd6e2966ab000d224395c85",
      "parents": [
        "e0d8b45eb4afd832309243f743fd59925d925750",
        "9efb727fd9527e2c76b58d5006715efd7a0164a8"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Oct 21 11:14:43 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Oct 21 11:14:43 2026 -0700"
      },
      "message": "Merge \"Bump bazel version to 9.3.0\""
    },
    {
      "commit": "8bd0a16180497358515ea28e92f8335d98300739",
      "tree": "3bcf8b1af8aad63198c0cde79d886acec845afc1",
      "parents": [
        "c6aa350acb774c7880683926f3cd4a41a673c9ff"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 21:09:50 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 21:20:37 2026 +0200"
      },
      "message": "Bazel: Align dependency versions with Gerrit\n\nUpdate Gitiles\u0027 Bazel dependency pins to match versions used by Gerrit:\n\n- asm: 9.9.1 -\u003e 9.10.1\n- byte-buddy: 1.18.12 -\u003e 1.18.14\n- commons-codec: 1.18.0 -\u003e 1.20.0\n- commons-lang3: 3.18.0 -\u003e 3.20.0\n- jetty: 12.1.12 -\u003e 12.1.13\n- mockito-core: 5.23.0 -\u003e 5.24.0\n- protobuf-java: 4.36.0 -\u003e 4.36.2\n- slf4j: 2.0.18 -\u003e 2.0.20\n\nChange-Id: I11489e8a587c8f1a8517feecf863a521a5ba4f6a\n"
    },
    {
      "commit": "e0d8b45eb4afd832309243f743fd59925d925750",
      "tree": "48f3256350630a3971849ea5441e3528b70e3b9e",
      "parents": [
        "78afdfca29aa36bb664bff7f8022f89d12fb2d2b",
        "e33961224bda4a542c76c925d0d242d636b20021"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Oct 20 12:11:47 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Oct 20 12:11:47 2026 -0700"
      },
      "message": "Merge \"Add a JavaScript test lane\""
    },
    {
      "commit": "c6aa350acb774c7880683926f3cd4a41a673c9ff",
      "tree": "a90ebd4948c2a9957a112456c28828ab4217f5ab",
      "parents": [
        "9efb727fd9527e2c76b58d5006715efd7a0164a8"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 20:54:58 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 21:06:29 2026 +0200"
      },
      "message": "Bump rules_jvm_external to 7.2 and drop root pinning workaround\n\nVersion 7.2 fixes pinned root precedence for layered Coursier installs.\nPreviously, the pinned policy was not applied during Coursier layering,\nand deduplication could retain both winning and losing versions.\nStrict duplicate checks could therefore fail before resolution.\n\nGitiles\u0027s root versions declared in MODULE.bazel now take precedence\nover versions contributed by JGit. Remove the per-artifact force_version\namendments. Retain the pinned conflict policy and strict\nduplicate-version checking.\n\nChange-Id: I18edd8fe955a393157e5c614e465562fd8f70622\n"
    },
    {
      "commit": "e33961224bda4a542c76c925d0d242d636b20021",
      "tree": "3c0880f5ae1c0a83ed2f6f09c779457c0d04c505",
      "parents": [
        "60fcf46fa383bd110bfd0037e0c321d37c0da92d"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 22 19:09:10 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Oct 20 19:05:49 2026 +0000"
      },
      "message": "Add a JavaScript test lane\n\nGitiles ships JavaScript to the browser and has no way to test any of\nit. This adds a way, with as little machinery as the problem allows.\n\nThe rule\n--------\n\nrules_nodejs is already a declared dependency, and its module registers\nhermetic Node toolchains for every supported platform, so\ntools/js_test.bzl needs no new dependency and no change to MODULE.bazel.\nThe rule takes Node from the runtime toolchain type, which Bazel selects\nfor the platform the test runs on, not from the toolchain type it\nselects for the platform build actions run on. A rule is needed at all\nbecause rules_nodejs 6.x ships toolchains only -- nodejs_test moved to\naspect_rules_js -- and because Bazel 9 removed the native sh_test that\ncould otherwise have wrapped a runner. What is left is small: a launcher\nthat invokes `node --test`, passes the `args` attribute and --test_arg\nthrough, and writes a JUnit report when Bazel asks for one.\n\nTests are written against node:test and node:assert, which are part of\nthe runtime. There is no package manager, no lockfile, no node_modules\nand no configuration file to drift.\n\nThey live in jstests/ at the root. javatests/ mirrors java/ package\nfor package, and the code under test is not Java and does not live\nunder java/ -- it is served out of resources/. Putting JavaScript\ntests inside that mirror would make the mirror mean two things. A\nsibling directory named for the language does not, and it reads the\nsame as javatests/ to anyone already used to the tree.\n\nThe smoke test\n--------------\n\nBuild glue that nothing exercises is build glue nobody knows is broken,\nso the rule arrives with a test of its own two promises: that the Node\nrunning the tests is the toolchain\u0027s and not one found on PATH, and that\na file listed in `data` is readable at its workspace-relative path even\nwhen it comes from another package. Both are environment-dependent, and\nneither is visible in the source of a test that merely passes.\n\nCoverage of the JavaScript itself follows in the changes that introduce\nthe code worth covering.\n\nVerification\n------------\n\n  bazel test //...\n\npasses and picks the new target up with no further configuration.\n\nThe lane also runs in CI rather than only locally. The gitiles-build job\nruns `bazelisk test //...` and fails on a non-zero, non-four exit, so a\nfailing JavaScript test fails the job. That was confirmed by observing\nVerified-1 on Change 646024, a throwaway change stacked on this one\nwhose only difference is one deliberately inverted assertion.\n\nBug: 564987520\nTAG\u003dagy\nCONV\u003d0b3e4b30-d902-4587-8bb2-ce23109d6955\nCONV\u003d60d1bfdf-34a4-460e-85ab-0f251e052543\nChange-Id: I11c6bbcf18c548f7507861be2fe71727796ca155\n"
    },
    {
      "commit": "9efb727fd9527e2c76b58d5006715efd7a0164a8",
      "tree": "b6ea807e2c2729e6fd329bd19e2445822c56a28c",
      "parents": [
        "78afdfca29aa36bb664bff7f8022f89d12fb2d2b"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 20:53:32 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Oct 20 21:05:36 2026 +0200"
      },
      "message": "Bump bazel version to 9.3.0\n\nChange-Id: I563610ae9f65f7077e4e11d067edfcd9e03213f0\n"
    },
    {
      "commit": "78afdfca29aa36bb664bff7f8022f89d12fb2d2b",
      "tree": "67908d95eb4a0d3797b6d204f846f64c2ad0e53f",
      "parents": [
        "2c8ddbd3a2829ed2082535f8778be9d4709d0f61"
      ],
      "author": {
        "name": "Thomas Dräbing",
        "email": "thomas.draebing@sap.com",
        "time": "Mon Oct 19 13:50:28 2026 +0200"
      },
      "committer": {
        "name": "Thomas Dräbing",
        "email": "thomas.draebing@sap.com",
        "time": "Mon Oct 19 13:50:33 2026 +0200"
      },
      "message": "Remove rbe_autoconfig\n\nThis should not be required anymore, since JGit doesn\u0027t use it\nanymore.\n\nChange-Id: Id1ede0657f7da3d11605efe0e26d67d0585ed46c\n"
    },
    {
      "commit": "2c8ddbd3a2829ed2082535f8778be9d4709d0f61",
      "tree": "ce5fb26a99578d7d8c53f75cadd69f3ac32de266",
      "parents": [
        "43221cc60541f30db8c80b70ce1f9666cabaf0ad"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Tue Oct 06 18:10:35 2026 +0000"
      },
      "committer": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Tue Oct 06 18:10:35 2026 +0000"
      },
      "message": "BlameServlet: Honor .git-blame-ignore-revs\n\nSkip bulk reformatting and other mechanical revisions during blame,\nattributing lines to the last meaningful change.\n\nIf .git-blame-ignore-revs exists at the root of the repository in the\ntree of the revision being blamed, its listed revisions are ignored\nwhen computing blame. If the file is missing, too large or cannot be\nread, blame ignores no revisions instead of failing the request.\n\nBoth the HTML and the JSON blame views honor the file. BlameCacheImpl\nsupports ignored revisions and includes them in its cache key.\n\nChange-Id: I9290cb31e3793e1e72e210b9c841a8508fe92046\n"
    },
    {
      "commit": "43221cc60541f30db8c80b70ce1f9666cabaf0ad",
      "tree": "e7ed7d33adeb18eadba5b402f9d90d5415cc1590",
      "parents": [
        "60fcf46fa383bd110bfd0037e0c321d37c0da92d"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Tue Oct 06 18:10:34 2026 +0000"
      },
      "committer": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Tue Oct 06 18:10:34 2026 +0000"
      },
      "message": "BlameCache: Add reader for .git-blame-ignore-revs files\n\nMany projects list bulk reformatting and other mechanical commits in a\n.git-blame-ignore-revs file at the root of the repository. git (through\nblame.ignoreRevsFile), GitHub and GitLab use this file to keep those\ncommits out of blame. BlameCache.get() already accepts a set of\nrevisions to ignore, but each caller would have to find, read and parse\nthe file itself.\n\nAdd IgnoreRevFileReader, which reads such a file from the tree of a\ncommit and returns an IgnoreRevsFile record containing the listed\nrevisions, ready to pass to BlameCache.get(). Keeping the reader in the\nblame cache library lets Gitiles and Gerrit share one implementation.\n\nThe format matches git: one full commit SHA-1 per line, \"#\" starts a\ncomment and blank lines are skipped. Unlike git, a malformed line is\nskipped instead of rejecting the whole file. Only regular files are\nread; a directory, symlink or submodule at the path is ignored.\n\nCallers may read the file on every blame request, so reading is\nbounded: files larger than 1 MiB are ignored and at most 20,000\nrevisions are used.\n\nThe BlameCache interface is unchanged.\n\nChange-Id: I381018685e49c327c6ba14a5974386535a311913\n"
    },
    {
      "commit": "60fcf46fa383bd110bfd0037e0c321d37c0da92d",
      "tree": "045ac211bcaf6266227f35627a62d5a3108a815f",
      "parents": [
        "0d4f8858d144beaaa7a85b35b5be9b528fdb8662",
        "4f81551cfbab5ff5f1f0258c592e7c6268a3d18c"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Sep 18 15:31:18 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Fri Sep 18 15:31:18 2026 -0700"
      },
      "message": "Merge \"Add paths_only projection to the recursive tree JSON API\""
    },
    {
      "commit": "4f81551cfbab5ff5f1f0258c592e7c6268a3d18c",
      "tree": "8569926167bab2a8a563d1884db5a660ccde7b87",
      "parents": [
        "77d36f6ff5f50316daa82c6753af2c3217859df4"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Sat Sep 12 00:51:36 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Sep 18 16:43:57 2026 +0000"
      },
      "message": "Add paths_only projection to the recursive tree JSON API\n\nAdds ?paths_only\u003d1 to the recursive tree listing, returning just the\nblob path names and omitting the per-entry mode, type and object ID.\nThis is the payload a path-oriented client such as a file finder\nneeds; the existing listing carries several times more bytes than\nsuch a client can use.\n\nMeasured against chromium/src at 506,237 blobs, ?recursive\u003d1\ntransfers 17.1 MB gzipped while ?recursive\u003d1\u0026paths_only\u003d1 transfers\n3.4 MB.\n\nThe listing is deliberately unbounded. A caller cannot distinguish a\npath omitted by a cap from a path that does not exist, so a partial\nlisting would be a correctness bug rather than a load mitigation. No\nbound is needed in any case: this is a strict projection of\n?recursive\u003d1, which is itself unbounded and considerably more\nexpensive to serve.\n\nBecause it is unbounded it is also streamed rather than collected.\nPathList.SOURCE_ADAPTER writes each path to the JsonWriter as the\nTreeWalk yields it, so serializing a tree of any size costs constant\nmemory; buffering chromium/src first would have held roughly 65 MB of\ntransient strings per concurrent request. PathList itself remains the\ndeclared response shape, and PathServletTest deserializes the\nstreamed bytes back into it, which is what holds the writer and the\nshape in agreement. The existing ?recursive\u003d1 listing still buffers\nand is not touched here.\n\nOn chromium/src this moves time to first byte from 234 ms to 4 ms\nwith total response time unchanged at 0.38 s. The bytes are\nidentical: the full 43 MB response compares equal before and after,\nas does the gzipped path.\n\nThe trade is that a mid-walk object store error now arrives as\ntruncated JSON rather than as an error status, because the response\nis already partly written. Callers parse the body, so truncation\nfails loudly rather than silently shortening the listing.\n\npaths_only requires recursive\u003d1 and rejects the request otherwise. A\nnon-tree target already yields 404 through WalkResult.recursivePath,\nso it needs no separate type check of its own.\n\nTAG\u003dagy\nCONV\u003d0b3e4b30-d902-4587-8bb2-ce23109d6955\nChange-Id: I010e8f8b1d8f6cd4bb28492375481653186bd44e\n"
    },
    {
      "commit": "0d4f8858d144beaaa7a85b35b5be9b528fdb8662",
      "tree": "faf41f54061cd1ca42812710b4385b6299ea653c",
      "parents": [
        "e5135e39eac0f5f3b91c7616f9131be32286a136"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Thu Sep 17 00:21:58 2026 +0000"
      },
      "committer": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Thu Sep 17 00:22:10 2026 +0000"
      },
      "message": "Render unchanged diff lines as grey instead of yellow\n\nIn Gitiles unified diff formatting (HtmlDiffFormatter), unchanged\ncontext lines (starting with \u0027 \u0027) are wrapped in \u003cspan class\u003d\"Diff-change\"\u003e.\nPreviously, dark mode support introduced --diff-change styled as #960\nin light mode and #fdd663 in dark mode, causing all unchanged lines\nin unified diffs to render as yellowish-brown / yellow instead of neutral\ngrey.\n\nUpdate --diff-change to #666 in light mode and #9aa0a6 in dark mode so\nthat unchanged context lines render in grey matching the secondary text\npalette.\n\nAdd unit tests to verify that unchanged lines are formatted with\nDiff-change and that --diff-change resolves to grey colors instead of\nyellow in base.css.\n\nBug: b/562476907\ngoogle-bug-id: 562476907\nTAG\u003dagy\nCONV\u003df9f02e8a-5730-4ba4-9c5c-f2415cbb4fd7\n\nChange-Id: Ie3c687845b707161f462b5e5cfe10ce481e9fa2b\n"
    },
    {
      "commit": "e5135e39eac0f5f3b91c7616f9131be32286a136",
      "tree": "2c8de116f87fbc4d53a1874cf9758c1a5bad1ec5",
      "parents": [
        "77d36f6ff5f50316daa82c6753af2c3217859df4"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Sep 14 19:19:58 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Wed Sep 16 08:29:36 2026 +0200"
      },
      "message": "Bazel: Package stamped JGit only in the Gitiles WAR\n\nJGit\u0027s default Bazel target is now non-stamped, so ordinary Gitiles\ncompile and test actions no longer depend on volatile workspace status.\n\nUpdate the JGit submodule to the change that adds the explicit\n//org.eclipse.jgit:jgit-stamped target, and use that target at the\ngitiles.war packaging boundary.\n\nExclude the non-stamped core JGit jar from the WAR so the packaged\napplication contains only the JGit jar with the Implementation-Version\nmanifest entry.\n\nFeature: Issue 561838827\nChange-Id: I3f90e2281004c1903dadc50919229982ef0ae48d\n"
    },
    {
      "commit": "77d36f6ff5f50316daa82c6753af2c3217859df4",
      "tree": "127145f63c4aaaec4503b87386b7fccf9dd42906",
      "parents": [
        "a9409d9529e6f2da746e126dba7a8f4126f01d3e",
        "f74576d007f6c5fbef5f6b226ab761626ddde2ea"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Wed Sep 09 12:50:13 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Sep 09 12:50:13 2026 -0700"
      },
      "message": "Merge \"Support code blocks within markdown tables\""
    },
    {
      "commit": "f74576d007f6c5fbef5f6b226ab761626ddde2ea",
      "tree": "377a832903fa3532fa0621cdf4bc3a8d3934879f",
      "parents": [
        "77be30bcac9ff7caaebd8d76634710b6c5a1f3b3"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Tue Sep 08 18:36:58 2026 +0000"
      },
      "committer": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Wed Sep 09 19:31:26 2026 +0000"
      },
      "message": "Support code blocks within markdown tables\n\nMarkdown tables (GFM tables) require rows to be specified on single\nlines, preventing standard fenced or indented code blocks within cells.\nDocumentation generators (such as Bazel Stardoc) generate code blocks\ninside table cells using \u003cpre\u003e\u003ccode\u003e...\u003c/code\u003e\u003c/pre\u003e (or \u003cpre\u003e...)\nwith HTML line breaks or entity-encoded newlines (\u0026#10;).\n\nPreviously, GitilesHtmlExtension ignored \u003cpre\u003e and \u003ccode\u003e inline\ntags, causing MarkdownToHtml to strip them and rendering code blocks as\nplain inline text with collapsed spaces and no code styling.\nAdditionally, when GitilesHtmlExtension unlinked replaced sibling nodes\n(such as closing anchor tags), the visitor traversal prematurely ended,\nskipping subsequent siblings in the same container.\n\nThis change adds:\n1. TableCodeBlock, a CustomNode representing code blocks in inline\n   contexts like table cells.\n2. Handling in GitilesHtmlExtension to scan \u003cpre\u003e / \u003ccode\u003e tags,\n   extract language attributes, preserve newlines and code formatting,\n   and convert them to TableCodeBlock (and FencedCodeBlock for block\n   HTML).\n3. Traversal in GitilesHtmlExtension.visitChildren that safely advances\n   past replaced sibling ranges without skipping siblings.\n4. Support for \u003ca id\u003d\"...\"\u003e anchors in addition to \u003ca name\u003d\"...\"\u003e.\n5. Visitor handling in MarkdownToHtml to render TableCodeBlock via\n   codeInPre(...).\n\nChange-Id: I75c6420f19537d6b9836294f58bcdc6405410192\n"
    },
    {
      "commit": "a9409d9529e6f2da746e126dba7a8f4126f01d3e",
      "tree": "77a462ada0efdf4f33b1b3ea22d899dd8dbffa8e",
      "parents": [
        "170bc1c13ef7b5c62624dc647c37ebb416402601",
        "875f0657401099b214f7d98b2ed5409e86d5e018"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Wed Sep 09 09:46:21 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Sep 09 09:46:21 2026 -0700"
      },
      "message": "Merge \"Include PRETTIFY_CSS_URL on repository index landing page when README is present\""
    },
    {
      "commit": "170bc1c13ef7b5c62624dc647c37ebb416402601",
      "tree": "0806c9da11093cf82d675bdbad56f1f4d9cd3913",
      "parents": [
        "77be30bcac9ff7caaebd8d76634710b6c5a1f3b3"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 08 20:29:50 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 08 16:09:52 2026 -0700"
      },
      "message": "Add dark mode theme support for Mermaid diagrams\n\nGitiles renders Mermaid diagrams server-side into SVG. Previously, all\nSVG shapes (nodes, subgraphs, edges, markers, text, and labels) used\nhardcoded light-mode hex values, causing them to display as glaring\nwhite rectangles with poor contrast in dark mode.\n\nThis change:\n- Defines CSS variables for Mermaid diagram elements (--mermaid-node-bg,\n  --mermaid-node-stroke, --mermaid-node-text, --mermaid-edge,\n  --mermaid-subgraph-bg, etc.) in doc.css under :root, light mode,\n  dark media query, and html[data-theme\u003d\"dark\"].\n- Adds semantic CSS classes (.mermaid-node, .mermaid-subgraph,\n  .mermaid-edge, .mermaid-arrow, .mermaid-shadow, etc.) to rendered SVG\n  elements in SimpleMermaidRenderer while preserving presentation\n  attribute fallbacks.\n- Selectively applies fill/stroke classes when custom style directives\n  are absent, preserving user-defined colors without stylesheet overrides.\n- Supports `color:` in `style` directives and computes background\n  luminance for custom-filled nodes (e.g. pastel database cylinders)\n  to preserve dark, readable text without stylesheet theme inversion.\n- Adds comprehensive unit tests in SimpleMermaidRendererTest covering\n  theme classes, custom styles, and text contrast on custom-filled shapes.\n\nChange-Id: Iac234a2dd627a06ecceb41c4fabbbbdd92233605\n"
    },
    {
      "commit": "875f0657401099b214f7d98b2ed5409e86d5e018",
      "tree": "87a05fd0985956983c14d8c37998caa94659e441",
      "parents": [
        "77be30bcac9ff7caaebd8d76634710b6c5a1f3b3"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 08 20:24:37 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Tue Sep 08 16:09:45 2026 -0700"
      },
      "message": "Include PRETTIFY_CSS_URL on repository index landing page when README is present\n\nRepositoryIndex.soy renders $readmeHtml when a root README.md exists,\nbut only passed DOC_CSS_URL to common.header. Pass PRETTIFY_CSS_URL as\nwell (matching PathDetail.soy and RevisionDetail.soy) so that fenced\ncode blocks in README.md on the repository landing page render with\nsyntax highlighting and theme variables.\n\nChange-Id: Ic7adc3a50e723d706135c275ea5e1efc6e525e83\n"
    },
    {
      "commit": "77be30bcac9ff7caaebd8d76634710b6c5a1f3b3",
      "tree": "ae45d96ae70fe2b30b833c86a6487d25061a2b6c",
      "parents": [
        "dc32d06d0d42bf9b6d9633fba630a100d28db874"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Thu Sep 03 07:35:38 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Thu Sep 03 07:47:26 2026 +0200"
      },
      "message": "Bazel: Align dependency versions with Gerrit\n\nUpdate Gitiles\u0027 Bazel dependency pins to match versions used by Gerrit.\n\nChange-Id: Id0370c5434f55e64ed57ae0e23f5222cc49c7b45\n"
    },
    {
      "commit": "dc32d06d0d42bf9b6d9633fba630a100d28db874",
      "tree": "cf25595b7fc6ab03c654416aa347254afc537946",
      "parents": [
        "4d1e38b64e6576eba470b09a1e3e463deda1dc11"
      ],
      "author": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Mon Aug 03 18:51:54 2026 +0530"
      },
      "committer": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Wed Sep 02 15:55:55 2026 +0200"
      },
      "message": "TreeSoyData: link tree pages to content search\n\nAdd a search link to repository tree pages. The link opens the +grep\nHTML view at the current revision and preserves the current directory,\nso searches from a subtree remain limited to that path.\n\nIssue: 376381593\nChange-Id: Ie1015b890b34edf23cf938651ca1edbe7e17cb72\nSigned-off-by: Devansh Varshney \u003cdevansh@libreoffice.org\u003e\n"
    },
    {
      "commit": "4d1e38b64e6576eba470b09a1e3e463deda1dc11",
      "tree": "751df94d218b92aeed310eb99ceb82c648bdfe30",
      "parents": [
        "968a50e4c1d36687e876a75ce50ec521d1bdbdf2"
      ],
      "author": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Wed Jun 24 21:32:04 2026 +0530"
      },
      "committer": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Wed Sep 02 15:54:48 2026 +0200"
      },
      "message": "GrepServlet: add HTML view for content search\n\nThe +grep endpoint currently supports JSON responses only. Add a\nbasic Soy-backed HTML page with a search form and linked file/line\nresults, reusing the existing bounded grep implementation.\n\nHTML requests without the s parameter show the search form without\nrunning a search. JSON behavior is unchanged and still requires s.\n\nIssue: 376381593\nChange-Id: Ie5b2a513d61fb4c44d6efba868ae6faa9aa866d0\nSigned-off-by: Devansh Varshney \u003cdevansh@libreoffice.org\u003e\n"
    },
    {
      "commit": "968a50e4c1d36687e876a75ce50ec521d1bdbdf2",
      "tree": "2350a5ccfafb581801c190482e3df98ffc080e80",
      "parents": [
        "5ce062876aab5328ca7d9e87bc4782fed3befa6a"
      ],
      "author": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Sat May 23 19:48:35 2026 +0530"
      },
      "committer": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Wed Sep 02 15:51:00 2026 +0200"
      },
      "message": "GrepServlet: add JSON content search endpoint\n\nAdd a +grep endpoint that searches blob contents at one revision using\nJGit TreeWalk and ObjectReader.\n\nThis starts the git grep part of the Gitiles search issue without adding\nan index or UI. The endpoint is JSON-only for now and keeps the search\nbounded by skipping binary or large blobs and stopping after a fixed\nnumber of matches.\n\nIssue: 376381593\nChange-Id: I351e81bb378d2a759c6b5c6d686199c598c6e886\n"
    },
    {
      "commit": "5ce062876aab5328ca7d9e87bc4782fed3befa6a",
      "tree": "abde9b36adfde766e51bdb9308734dac1345ac3f",
      "parents": [
        "48e507c9a527e955a80faa6610f5eefa292b9059",
        "0b8ab6c71efb23845cfab57a588ac902bc6ad5aa"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Tue Sep 01 07:55:14 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Sep 01 07:55:14 2026 -0700"
      },
      "message": "Merge changes from topic \"gitiles-source-submodule\"\n\n* changes:\n  Mark war-provided libraries as neverlink inside the Gerrit tree\n  Make webassets packaging repo-path independent\n"
    },
    {
      "commit": "48e507c9a527e955a80faa6610f5eefa292b9059",
      "tree": "b696f3fdc3120bbe253878c18dbda2edf187bf8c",
      "parents": [
        "7426564caf1f66604769286e306aba681046fe6d"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Sat Aug 29 21:44:25 2026 +0000"
      },
      "committer": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Mon Aug 31 17:08:38 2026 +0000"
      },
      "message": "Address all static analysis findings from internal import\n\nFix all static analysis findings identified during internal import across\n12 affected files in Gitiles:\n\n- Documentation: Add missing Javadoc comments to public classes,\n  interfaces, builder classes, and enums.\n- Language \u0026 Syntax Modernization: Convert multiline strings to Java text\n  blocks, convert statement switches to arrow expression switches,\n  adopt pattern matching for instanceof and switches, and group method\n  overloads together without interleaving helper methods.\n- Deprecation Resolutions: Migrate from deprecated PersonIdent constructor\n  to Instant/ZoneId variants, replace OrderedList.getStartNumber() with\n  getMarkerStartNumber(), and replace Splitter.onPattern() with\n  Splitter.on(Pattern.compile(...)).\n- Guava \u0026 Truth Assertions: Adopt ImmutableList/ImmutableSet where\n  appropriate, use buildOrThrow(), use computeIfAbsent(), and modernize\n  Truth assertions with isPresent(), hasSize(), and isEmpty().\n- Style \u0026 Cleanliness: Enforce lowerCamelCase identifier names, bracket\n  single-line control flow statements, split multi-variable declarations,\n  avoid Yoda conditions, and make utility class constructor private.\n\nChange-Id: Ia76966182374aef1c8cf6fb8d207567f4615c8f1\n"
    },
    {
      "commit": "0b8ab6c71efb23845cfab57a588ac902bc6ad5aa",
      "tree": "1fcf3a97f4db6cdf61bc5e48b7bcb00431729530",
      "parents": [
        "8bb664ebb09f271a6cc2996e401d8ae9ebdef478"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 31 15:28:02 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 31 15:42:21 2026 +0200"
      },
      "message": "Mark war-provided libraries as neverlink inside the Gerrit tree\n\nWhen Gitiles is built as a Gerrit plugin from source, the plugin jar was\npackaging the full transitive closure of libraries that Gerrit already\nships in gerrit.war (Guava, Gson, Soy, ICU4J, JGit, commons-lang3/io,\ntext/reload4j/flogger, ASM, html-types, ...). That inflated gitiles.jar\nto ~34 MB of classes that are already on the plugin\u0027s runtime classpath\nvia gerrit.war -- pure duplication.\n\nIntroduce lib/provided.bzl with a provided_java_library() wrapper: a\nsingle java_library whose neverlink attribute is a select() on the\nexisting @com_googlesource_gerrit_bazlets//flags:in_gerrit_tree flag.\n\nGerrit sets that flag in its .bazelrc; standalone Gitiles never does, so\nit defaults to false.\n\n* Standalone (flag false): the libraries are linked exactly as before\n  and bundled into gitiles.war -- no behavioral change, zero regression\n  risk.\n* In the Gerrit tree (flag true): they become neverlink, so their\n  classes are dropped from the plugin\u0027s runtime closure and are not\n  re-packaged into gitiles.jar. Gerrit already provides them.\n\nOnly libraries that gerrit.war actually ships are converted (lib/BUILD,\nlib/soy, lib/flogger). Gitiles-only libraries (the commonmark family,\njava-prettify and autolink) stay linked in both configurations.\n\nAlternative considered: exclude the same closure with deploy_env from\nwithin the Gerrit plugin\u0027s BUILD (the pattern used by other plugins). We\ndefer that here because in_gerrit_tree is more principled -- Gitiles\nitself declares \"these are provided when I\u0027m built inside Gerrit,\" so\nthe knowledge lives with the dependency declarations instead of being\nduplicated as an exclusion list in Gerrit\u0027s plugin wiring. deploy_env\nremains a viable follow-up if we later prefer to keep this decision\nentirely on the Gerrit side.\n\nChange-Id: I4f12d3a4be71a89df8830ecc872877f869997a8c\n"
    },
    {
      "commit": "8bb664ebb09f271a6cc2996e401d8ae9ebdef478",
      "tree": "f41dcf9a2e7da5f400abb8cd45f635ae8700ebb6",
      "parents": [
        "7426564caf1f66604769286e306aba681046fe6d"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 31 12:40:05 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 31 12:55:58 2026 +0200"
      },
      "message": "Make webassets packaging repo-path independent\n\nGitiles\u0027 webassets genrule copied resources by stripping one\npath component from the source paths. That works when Gitiles is\nthe main repository, where resource paths start with resources/...,\nbut not when another Bzlmod build consumes Gitiles as a local\nexternal repository.\n\nIn that case Bazel passes source paths below its external repository\ntree, so stripping one component leaves the wrong layout and the rule\ncannot cd into com/google/gitiles.\n\nCopy each source under the path segment after\nresources/com/google/gitiles/ instead. This preserves the existing\narchive layout for standalone Gitiles and allows Gerrit to\nsource-build Gitiles from a modules/gitiles submodule while supplying\nits own JGit and java-prettify repositories.\n\nChange-Id: Iec8dbd92411d0fd5a5669035dadc6da231242fb8\n"
    },
    {
      "commit": "7426564caf1f66604769286e306aba681046fe6d",
      "tree": "f09c3ff07eac628a9a0aef57132e8adde6ea0a6b",
      "parents": [
        "cbbc9e3bd4c59ba0cf46c543b3d2d426488ac7ad",
        "3c524d03cb015c77f03c15e2d64ee2ddf63b1cfe"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Mon Aug 31 03:16:30 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Mon Aug 31 03:16:30 2026 -0700"
      },
      "message": "Merge \"Fix broken tar.xz archive download\""
    },
    {
      "commit": "cbbc9e3bd4c59ba0cf46c543b3d2d426488ac7ad",
      "tree": "a8615540a412794434d068252b4ad29a222babe0",
      "parents": [
        "9b632a9d4ab07242ffbce16ea3ae1c71ad859b32",
        "259f87e50f34d011ee3a93f822d6150c3bbc7111"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Mon Aug 31 03:16:17 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Mon Aug 31 03:16:17 2026 -0700"
      },
      "message": "Merge \"Switch to Java 25\""
    },
    {
      "commit": "9b632a9d4ab07242ffbce16ea3ae1c71ad859b32",
      "tree": "4480a4d74749ff137819062d29f2d3b9b61fc652",
      "parents": [
        "55c1f19761e1fb594e080b9e135bd1eda0e17ca9",
        "843e958b9395ab7533a16bae6971f72baf06a7c5"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Mon Aug 31 03:09:34 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Mon Aug 31 03:09:34 2026 -0700"
      },
      "message": "Merge \"Bump protobuf to 36.0 and rules_java to 9.9.0\""
    },
    {
      "commit": "3c524d03cb015c77f03c15e2d64ee2ddf63b1cfe",
      "tree": "8160adc67bd994faefcdc5f0e9681044d49c7218",
      "parents": [
        "259f87e50f34d011ee3a93f822d6150c3bbc7111"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Aug 29 09:54:18 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 30 22:50:15 2026 +0200"
      },
      "message": "Fix broken tar.xz archive download\n\nGitiles advertises the tar.xz archive format (ArchiveFormat.TXZ), but\nthe generated gitiles.war did not package org.tukaani.xz. JGit\u0027s Bazel\njgit-archive target failed to declare it (the //lib:xz target was\ntestonly), so it was never pulled transitively. Requesting a .tar.xz\narchive therefore failed at runtime with NoClassDefFoundError:\norg/tukaani/xz/..., and no test caught it because the only archive test\nexercised tar.gz.\n\nBump the JGit submodule to pick up the fix that declares org.tukaani.xz\nas a dependency of the archive library, and add a regression test that\nstreams a .tar.xz archive.\n\nDepends-On: https://eclipse.gerrithub.io/c/eclipse-jgit/jgit/+/1247298\nChange-Id: I1cd020dc4c4d0e3059d7f3b57ac2771f6037c50d\n"
    },
    {
      "commit": "55c1f19761e1fb594e080b9e135bd1eda0e17ca9",
      "tree": "0a00423416fc67f671ff4656589e8c8e0c96c7a2",
      "parents": [
        "2bdd88de3d6ccd5d5cf7c773b2a47d6b788069f8"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Aug 29 16:27:07 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 30 07:52:15 2026 +0200"
      },
      "message": "Format Mermaid SVG numbers with Locale.ROOT\n\nSimpleMermaidRenderer built SVG coordinates with\nString.format(\"%.1f\", ...) under the default locale, so on a JVM whose\nregion uses a comma decimal separator (e.g. de_DE) it emitted \"28,0\"\ninstead of \"28.0\". SVG requires \u0027.\u0027 as the decimal separator, so that\nmarkup is invalid; the renderer tests, which reparse coordinates with\nDouble.parseDouble, throw NumberFormatException for the same reason.\n\nPass Locale.ROOT to every String.format call so numeric output is\nalways \u0027.\u0027-separated and independent of the machine locale.\n\nAlternative considered: forcing an en-US locale from .bazelrc or\noverriding the JVM/system default locale. That does not make the\nrenderer correct on its own, and it still leaves production output\ndependent on process-wide locale configuration. The formatting point\nthat emits SVG numbers should declare its locale explicitly instead.\n\nChange-Id: Ia493dd426d3a1dffb5b7059df69c5a0001b1dab1\n"
    },
    {
      "commit": "259f87e50f34d011ee3a93f822d6150c3bbc7111",
      "tree": "6061b43dcf836603e0bf18e735cb62020bd66208",
      "parents": [
        "843e958b9395ab7533a16bae6971f72baf06a7c5"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Aug 28 18:37:30 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Aug 29 10:54:58 2026 +0200"
      },
      "message": "Switch to Java 25\n\nBump the build and tool Java language/runtime versions from 21 to 25 and\nretarget the Error Prone default_java_toolchain at remotejdk_25.\n\nNo sun.misc.Unsafe workaround is needed: rules_java 9.9.0 (the parent\ncommit) bundles protobuf 36.0 for JavaBuilder and turbine, which no\nlonger call the sun.misc.Unsafe memory-access methods that required\n--sun-misc-unsafe-memory-access\u003dallow on JDK 24+. The JDK \u003c\u003d 21\n-XDaddTypeAnnotationsToSymbol workaround is also dropped, as it was a\nno-op on JDK 22+.\n\nAlso bump the Eclipse compiler compliance, source and target levels in\n.settings/org.eclipse.jdt.core.prefs from 1.8 to 25. That file dated\nback to when Gitiles was a Java 8 project and was never updated, so\nEclipse rejected Java 9+ syntax (e.g. the diamond operator on anonymous\nclasses) in imported sources even though the JRE container is JavaSE-25.\n\nChange-Id: Ic4765288722a9f3b2f4413cc9708d9a024f39c61\n"
    },
    {
      "commit": "843e958b9395ab7533a16bae6971f72baf06a7c5",
      "tree": "29869cf13a056efb05ab895e1cb27bdd462d303e",
      "parents": [
        "2bdd88de3d6ccd5d5cf7c773b2a47d6b788069f8"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Aug 28 18:34:39 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat Aug 29 09:59:13 2026 +0200"
      },
      "message": "Bump protobuf to 36.0 and rules_java to 9.9.0\n\nrules_java 9.9.0 bundles protobuf 36.0 for JavaBuilder and turbine,\nwhich no longer relies on the sun.misc.Unsafe memory-access methods that\nrequired --sun-misc-unsafe-memory-access\u003dallow on JDK 24+\n(protocolbuffers/protobuf#20760). Bumping it lets the upcoming Java 25\nswitch drop that workaround from the toolchain definition.\n\nAlso bump the Maven protobuf-java runtime artifact from 4.33.4 to 4.36.0\nto keep the protobuf version consistent, and re-pin\nexternal_deps.lock.json.\n\nChange-Id: Ib5132c5927e2ff720523c9b6a49131e2016211ed\n"
    },
    {
      "commit": "2bdd88de3d6ccd5d5cf7c773b2a47d6b788069f8",
      "tree": "c419a671c3e4c8d934e2d15eccbf9af61a329aad",
      "parents": [
        "2b3a9a56658c249ec037e7e32452360b18386beb",
        "33ba186227eb691fae79ce9423290f7f6b58efab"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Aug 28 19:12:25 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Fri Aug 28 19:12:25 2026 -0700"
      },
      "message": "Merge \"Add canonical cross-browser favicon package\""
    },
    {
      "commit": "2b3a9a56658c249ec037e7e32452360b18386beb",
      "tree": "737c1f3a6ff812dc6ac5fd7c87befa5213f3d514",
      "parents": [
        "4534419144eef40f9fb50ead5be28d65d7963396",
        "4b8a8ad410c19cd16a759600304e005142753320"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Fri Aug 28 16:21:29 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Fri Aug 28 16:21:29 2026 -0700"
      },
      "message": "Merge \"BlameCache: Support ignored revisions in get()\""
    },
    {
      "commit": "33ba186227eb691fae79ce9423290f7f6b58efab",
      "tree": "d75ff70020e0a0451fee06c8ef8f615c3ed6f574",
      "parents": [
        "4d478d36f4f3b56695a2250496e95ec83614d239"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Thu Aug 27 14:28:19 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Aug 28 19:50:17 2026 +0000"
      },
      "message": "Add canonical cross-browser favicon package\n\nIntroduce a complete, brand-neutral Git favicon set based on the\nGit coral-orange squircle with white branching commit nodes.\n\nIncludes:\n- favicon.svg: Lossless scalable vector for modern browsers\n- favicon.ico: Multi-resolution binary (16x16, 32x32, 48x48)\n- favicon-32x32.png \u0026 favicon-16x16.png: High-DPI and standard PNGs\n- apple-touch-icon.png: 180x180 for iOS/macOS Safari bookmarks\n- android-chrome-192x192.png \u0026 android-chrome-512x512.png: High-resolution assets\n\nUpdates Renderer.java and HTML head templates in Common.soy and Doc.soy\nto declare the canonical favicon links. Adds comprehensive servlet unit\ntests in FaviconTest.\n\nChange-Id: Ia4e17c7a17da2813a767a0d4f000afc5c6ccc284\n"
    },
    {
      "commit": "4534419144eef40f9fb50ead5be28d65d7963396",
      "tree": "673d37bd81f961c5ef62b7e961cc87fe29e1008d",
      "parents": [
        "4d478d36f4f3b56695a2250496e95ec83614d239"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Aug 28 16:57:15 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Aug 28 17:58:43 2026 +0000"
      },
      "message": "Fix CheckReturnValue and remove indirect Doubles dependency in Mermaid\n\nWhen building Gitiles with strict ErrorProne checks during Google3\nimport (CL 972611210), compilation failed due to:\n\n1. In SimpleMermaidRenderer.CharScanner: next(), consume(String), and\n   consumeIgnoreCase(String) return values were ignored when advancing\n   tokens or skipping delimiters, triggering 64 CheckReturnValue compiler\n   errors. Refactor CharScanner to provide void advance(), void skip(String),\n   and void skipIgnoreCase(String) for token skipping, and boolean\n   tryConsume(String) / tryConsumeIgnoreCase(String) for the few sites\n   where the return value is tested. This eliminates unused return values\n   and avoids unnecessary CanIgnoreReturnValue annotations.\n2. In SimpleMermaidRendererTest:\n   - render(code5) return value was unused; assert that the returned\n     SvgDoc is not null.\n   - Doubles.tryParse(part) pulled in com.google.common.primitives.Doubles\n     as an indirect dependency; replace with inline Double.parseDouble(part)\n     in a try/catch block.\n\nChange-Id: I32de733a3bae5ba75a50d9f74af4c21084940f79\n"
    },
    {
      "commit": "4d478d36f4f3b56695a2250496e95ec83614d239",
      "tree": "199a4234064ca6d1fb546d2ca57f868eb85b0601",
      "parents": [
        "da05810b7b9c1424aa73832f230396dfd70cbe01",
        "2f1a119814bfeabaa1f80ebf87c79d71a984a437"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Fri Aug 28 08:22:52 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Fri Aug 28 08:22:52 2026 -0700"
      },
      "message": "Merge \"Add dark mode support to Gitiles\""
    },
    {
      "commit": "4b8a8ad410c19cd16a759600304e005142753320",
      "tree": "a9794f6211375a1f4c97a9428bfb6fcdc2f9f4cb",
      "parents": [
        "da05810b7b9c1424aa73832f230396dfd70cbe01"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Fri Aug 28 14:47:53 2026 +0000"
      },
      "committer": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Fri Aug 28 08:14:41 2026 -0700"
      },
      "message": "BlameCache: Support ignored revisions in get()\n\nAdd 4-argument get(repo, commitId, path, ignoreIds) to BlameCache and\nBlameCacheImpl.\n\nWhen ignoreIds is non-empty, BlameCacheImpl passes the set of ObjectIds\nto BlameGenerator.setIgnoreRevs() so that blame calculations skip the\nspecified commits and attribute the affected lines to parent commits.\n\nCache entries are stored under BlameCacheImpl.Key(commitId, path,\nignoreIds), guaranteeing complete isolation from standard unignored\nblame cache entries.\n\nChange-Id: I595f0fc62e65a2a7d26d3ed5af1d6fccb6e0742d\n"
    },
    {
      "commit": "2f1a119814bfeabaa1f80ebf87c79d71a984a437",
      "tree": "2b2f37ebf02fb120cfc183d8d793beae79e2b4bf",
      "parents": [
        "658b32a8187d43e9a8429183719c580baa084544"
      ],
      "author": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Wed Aug 26 22:43:53 2026 +0000"
      },
      "committer": {
        "name": "George Engelbrecht",
        "email": "engeg@google.com",
        "time": "Thu Aug 27 23:55:05 2026 +0000"
      },
      "message": "Add dark mode support to Gitiles\n\nImplement CSS custom properties across core UI, markdown docs, and\nsyntax highlighting stylesheets. Add a zero-FOUC theme initialization\nscript in \u003chead\u003e and an interactive toggle in the header that persists\npreference across sessions via localStorage and cookies.\n\nTested:\n- Added comprehensive unit tests in ThemeTest covering all HTML routes.\n- Ran all Gitiles servlet unit tests and downstream Gerrit HTTP tests.\n\nChange-Id: I2b8bf5dacd986710a62b61df87550d776ca20290\n"
    },
    {
      "commit": "da05810b7b9c1424aa73832f230396dfd70cbe01",
      "tree": "38aa7626feeeadeb75ee46737f77d527acd847f5",
      "parents": [
        "1cb80ac39b2c7e088d61f5b7649ad1a869c18257",
        "305570c765708225b2c19880861473bb5b6ef177"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Thu Aug 27 15:47:53 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Thu Aug 27 15:47:53 2026 -0700"
      },
      "message": "Merge \"Add server-side Mermaid flowchart diagram rendering to Markdown\""
    },
    {
      "commit": "1cb80ac39b2c7e088d61f5b7649ad1a869c18257",
      "tree": "006d2a2ea3841c5093a94144f92685e34919b365",
      "parents": [
        "2d56b485bc542d0ffc80663b9581f31662addcc9"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Thu Aug 27 17:14:23 2026 +0000"
      },
      "committer": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Thu Aug 27 11:42:33 2026 -0700"
      },
      "message": "BlameCacheImpl: Add ignoreIds to cache Key\n\nAdd ignored revisions (ignoreIds) to BlameCacheImpl.Key, including\ngetters, equals(), hashCode(), and toString() (formatting a\ndeterministically sorted list of revision SHAs).\n\nThis allows downstream implementations to safely update their key\nformatting and storage before introducing the get(..., ignoreIds) API.\n\nChange-Id: Ib2a71ea0e6c09babaf76354a50bceb2078e559a2\n"
    },
    {
      "commit": "2d56b485bc542d0ffc80663b9581f31662addcc9",
      "tree": "f14f17049e9d9b81d0b40f389106569f4fac86d0",
      "parents": [
        "658b32a8187d43e9a8429183719c580baa084544"
      ],
      "author": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Thu Aug 27 17:14:06 2026 +0000"
      },
      "committer": {
        "name": "Ravi Mistry",
        "email": "rmistry@google.com",
        "time": "Thu Aug 27 10:16:14 2026 -0700"
      },
      "message": "Update JGit submodule to fe01de69f\n\nUpdate JGit submodule to fe01de69f\n(https://gerrit.googlesource.com/jgit/+/fe01de69f44611049b664188652bf9e340f0b1f8)\nwhich adds support for ignoring revisions during blame traversal in\nBlameGenerator\n(https://review.gerrithub.io/c/eclipse-jgit/jgit/+/1228557), and repin\nexternal dependencies.\n\nChange-Id: I6da38beb51f03b72f5c19a60a95684d4d4ed7edd\n"
    },
    {
      "commit": "305570c765708225b2c19880861473bb5b6ef177",
      "tree": "8b469807116441792b9576cc715a0d0ec59844db",
      "parents": [
        "658b32a8187d43e9a8429183719c580baa084544"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Wed Aug 26 01:36:26 2026 +0000"
      },
      "committer": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Wed Aug 26 23:17:57 2026 +0000"
      },
      "message": "Add server-side Mermaid flowchart diagram rendering to Markdown\n\nIntroduce server-side SVG rendering for Mermaid flowcharts and graph\ndiagrams (both \u0027graph\u0027 and \u0027flowchart\u0027 in TD, TB, LR, RL, BT orientations)\nembedded in Gitiles Markdown documents.\n\nExample rendering:\nhttps://imgur.com/a/JMuzziu\n\nKey components:\n- AST Parser: Token-based recursive scanner handling all standard node\n  shapes (rectangle, rounded, stadium, subroutine, cylinder, circle,\n  hexagon, diamond, asymmetric flag), full edge matrix (solid, dashed,\n  thick, arrows, lines, bidirectional), infix/inline edge labels,\n  multi-node ampersand chaining (e.g. A \u0026 B --\u003e C \u0026 D), and nested\n  subgraphs with direction overrides.\n- Layered Layout Engine: Sugiyama-inspired DAG layout supporting Coffman-\n  Graham layering, barycentric crossing minimization, connected\n  component partitioning for isolated subgraphs, lateral edge routing,\n  and cycle-breaking loopback paths.\n- Security \u0026 Hardening: Clean AST-to-SVG emission ensuring zero HTML/DOM\n  injection, strictly escaping XML text/tspans, eliminating raw\n  \u003cforeignObject\u003e/\u003ciframe\u003e elements, and ignoring malicious click/style\n  directives.\n- Fallback Mechanism: Safely falls back to raw \u003cpre class\u003d\"mermaid\"\u003e\n  when diagrams are unsupported or malformed.\n- Test Coverage: 100% test line coverage with comprehensive unit,\n  integration, and security attack test suites.\n\nBug: b/490107924\nChange-Id: I72ddb3eb67c2513c22444b62581ebd04aaa4afd2\nTAG\u003dagy\nCONV\u003d85f7f320-abab-4b77-9892-70da2847776b\n"
    },
    {
      "commit": "658b32a8187d43e9a8429183719c580baa084544",
      "tree": "7c2779aff261059f788d004c231a110fd9248df0",
      "parents": [
        "367b7b18a201937ddc44846c1012e32abdb26102"
      ],
      "author": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Mon Aug 24 18:26:19 2026 +0000"
      },
      "committer": {
        "name": "Sam Saccone",
        "email": "samccone@google.com",
        "time": "Mon Aug 24 18:26:19 2026 +0000"
      },
      "message": "Allow formatting parameters (?format\u003dTEXT, ?format\u003dJSON) on markdown files\n\nWhen requesting a markdown file path via the auto command `+` (e.g.\n`/repo/+/master/README.md?format\u003dTEXT`), ViewFilter previously\nunconditionally mapped all `.md` paths to `GitilesView.doc()`\n(`DocServlet`). Because `DocServlet` only handles HTML rendering and\ndoes not support TEXT (base64) or JSON representations, requesting\n`format\u003dTEXT` or `format\u003dJSON` resulted in an HTTP 400\n(UNSUPPORTED_RESPONSE_FORMAT) error.\n\nThis change updates `ViewFilter.parseAutoCommand` to check the requested\nformat. It only routes to `GitilesView.doc()` when the format is HTML or\nDEFAULT (the standard web browser view). For TEXT or JSON requests, it\npreserves `GitilesView.path()` so `PathServlet` serves the base64 or\nJSON representation as expected, matching the behavior for non-markdown\nfiles and the `+show` command.\n\nChange-Id: Idd02ce77dba03bcc922983821ee3f55c37bfa878\n"
    },
    {
      "commit": "367b7b18a201937ddc44846c1012e32abdb26102",
      "tree": "c7ba1702775c3d6eea355e95e487c786e2cb6c76",
      "parents": [
        "04e04f2b1bd79a555851082d32d592adbf0bd503"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 16 13:46:57 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Wed Aug 19 10:06:49 2026 -0700"
      },
      "message": "Bump rules_java to 9.8.0\n\nUpdate rules_java from 9.3.0 to 9.8.0. The newer java_tools ship a\nnewer Error Prone that flags pre-existing violations; those are\naddressed in the preceding changes. Gitiles already runs its Java\ntools on JDK 21, so no toolchain runtime change is needed.\n\nChange-Id: Ibb3ce0a5fb3bb30b3b69b0aa39c9377a4508e56c\n"
    },
    {
      "commit": "04e04f2b1bd79a555851082d32d592adbf0bd503",
      "tree": "dcc193dafa8ae8efdebfee87d7344bf67f159ab0",
      "parents": [
        "8a17bff719ee0e59353fb376215186a5356b4db1",
        "4db80928643072a95f17b71153a0f0e69cdb9d08"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david.ostrovsky@gmail.com",
        "time": "Wed Aug 19 10:06:28 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Aug 19 10:06:28 2026 -0700"
      },
      "message": "Merge \"Fix ReferenceEquality Error Prone errors\""
    },
    {
      "commit": "8a17bff719ee0e59353fb376215186a5356b4db1",
      "tree": "b44f2476e9479e1d8f982b0dd1ce6189d53b0ec2",
      "parents": [
        "67c7844b7f7f228229c037cf297296874f50692c"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 16 13:46:57 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 17 21:12:00 2026 -0700"
      },
      "message": "Avoid NullArgumentForNonNullParameter false positive via javac flag\n\nrules_java 9.8.0\u0027s newer Error Prone flags idiomatic Guava calls\n(Iterables.getFirst(x, null); ToStringHelper.add(name, nullable) under\nomitNullValues) as NullArgumentForNonNullParameter. This is a false\npositive rooted in OpenJDK JDK-8225377: before JDK 22, javac does not\nattach type-use annotations loaded from classpath class files to their\nsymbols, so Error Prone cannot see Guava\u0027s @ParametricNullness /\nJSpecify @Nullable and treats the parameter as non-null. It reproduces\non the JDK 21 this repo targets (including the bundled 21.0.12) but not\non JDK 22+.\n\nSee this issue for more details:\nhttps://github.com/bazelbuild/bazel/issues/30743\n\nA TODO tracks removing the javacopt once JavaBuilder passes it upstream.\n\nChange-Id: I3fce51b27ada2a440b55ccb94b2d2e40ffff57c0\n"
    },
    {
      "commit": "67c7844b7f7f228229c037cf297296874f50692c",
      "tree": "b33cae9060d6acec6135350920d8fce21aa7a7b1",
      "parents": [
        "cbc954b18ae5a4677ee925688a3eb5dfb59012f3"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 16 13:46:57 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Aug 17 21:11:31 2026 -0700"
      },
      "message": "Fix JdkObsolete Error Prone errors by using Charset overloads\n\nThe newer Error Prone shipped with rules_java 9.8.0 reports\nJdkObsolete for the String-charset-name JDK APIs. Switch to the\nCharset overloads (URLEncoder/URLDecoder, new String,\nOutputStreamWriter, ByteArrayOutputStream.toString). Where the\ncharset is a StandardCharsets constant, .name()/the literal is\nreplaced by the constant; where it is dynamic, Charset.forName(...)\nis used. Now-dead UnsupportedEncodingException try/catch blocks and\nimports are removed.\n\nChange-Id: Ie4e8025bb463da9ebe80fe70af75bf4fdc9a9c4b\n"
    },
    {
      "commit": "4db80928643072a95f17b71153a0f0e69cdb9d08",
      "tree": "905f1d7bfdb9a272be50d169e25d5f023c4312cf",
      "parents": [
        "cbc954b18ae5a4677ee925688a3eb5dfb59012f3"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 16 13:46:56 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sun Aug 16 22:47:14 2026 +0200"
      },
      "message": "Fix ReferenceEquality Error Prone errors\n\nThe newer Error Prone shipped with rules_java 9.8.0 flags several\nreference-equality comparisons. Fix them following the conventions\nalready used in this codebase, keeping the check at ERROR:\n\n  - JGit FileMode comparisons use value equality (FileMode.X.equals),\n    matching PathServlet/ReadmeHelper: ArchiveServlet, PathServlet,\n    ImageLoader.\n  - commonmark AST nodes are intentionally compared by identity, so\n    suppress the check on those methods (as Revision already does):\n    MarkdownToHtml, NamedAnchorExtension, SmartQuotedExtension.\n\nChange-Id: Ie852ea8484717501e6b8302c71588a2bb45a230b\n"
    },
    {
      "commit": "cbc954b18ae5a4677ee925688a3eb5dfb59012f3",
      "tree": "85f52bdb669a82617492e0d4ad1a2b3a1dcffd6c",
      "parents": [
        "acd743facc8e9520173804d2a4ec93382fdc628f",
        "2151f53d2091f62c2e5544f36dcebebc139b9f5e"
      ],
      "author": {
        "name": "Nasser Grainawi",
        "email": "nasser.grainawi@oss.qualcomm.com",
        "time": "Wed Jul 29 10:08:15 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Jul 29 10:08:15 2026 -0700"
      },
      "message": "Merge \"Cache sorted ref lists in CommitData.getRefsById\""
    },
    {
      "commit": "acd743facc8e9520173804d2a4ec93382fdc628f",
      "tree": "d38ec78663e0cc5da66122334bccdf7070efe921",
      "parents": [
        "f2bd57f33548e0878d3eb4dcfde3e853871666e6",
        "e4dcf3a240d76c85134dee06e593e2fdc1c280b6"
      ],
      "author": {
        "name": "Nasser Grainawi",
        "email": "nasser.grainawi@oss.qualcomm.com",
        "time": "Tue Jul 28 14:21:04 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Jul 28 14:21:04 2026 -0700"
      },
      "message": "Merge \"Add markdown.frontmatter option to strip YAML front matter\""
    },
    {
      "commit": "2151f53d2091f62c2e5544f36dcebebc139b9f5e",
      "tree": "658ed72839cb3033e665095d3af54f7025d25807",
      "parents": [
        "e7402f5c7ea414b3d86df68e26aaf85ed2083908"
      ],
      "author": {
        "name": "Ajay Gupta",
        "email": "ajagup@qti.qualcomm.com",
        "time": "Wed Jul 22 19:48:46 2026 +0530"
      },
      "committer": {
        "name": "Ajay Gupta",
        "email": "ajagup@qti.qualcomm.com",
        "time": "Thu Jul 23 07:36:16 2026 +0530"
      },
      "message": "Cache sorted ref lists in CommitData.getRefsById\n\ngetRefsById re-sorted the refs pointing at a commit on every call,\neven though the per-prefix index it looks them up in is built once\nand never changes afterward. Sort each id\u0027s refs when the index is\nbuilt instead, so repeated lookups for the same prefix reuse the\nalready-sorted list rather than re-sorting on every commit rendered.\n\nChange-Id: I8027c900e97f471dd6ba91b6b1b2ceeaca0afb2b\n"
    },
    {
      "commit": "f2bd57f33548e0878d3eb4dcfde3e853871666e6",
      "tree": "0041163bb6dcae60c4fe6b2e20f2159e64701c1b",
      "parents": [
        "bcbf5b005af0c682b2e5f393324569dd2b3843ab",
        "e7402f5c7ea414b3d86df68e26aaf85ed2083908"
      ],
      "author": {
        "name": "Nasser Grainawi",
        "email": "nasser.grainawi@oss.qualcomm.com",
        "time": "Wed Jul 22 08:00:33 2026 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Jul 22 08:00:33 2026 -0700"
      },
      "message": "Merge \"Avoid scanning every ref to render the commit log\""
    },
    {
      "commit": "e4dcf3a240d76c85134dee06e593e2fdc1c280b6",
      "tree": "6630398f48bc544cdf53e2d994180970eec1ba08",
      "parents": [
        "bcbf5b005af0c682b2e5f393324569dd2b3843ab"
      ],
      "author": {
        "name": "Alexander Daryin",
        "email": "shurick@google.com",
        "time": "Tue Jul 21 22:37:13 2026 +0000"
      },
      "committer": {
        "name": "Alexander Daryin",
        "email": "shurick@google.com",
        "time": "Tue Jul 21 22:37:13 2026 +0000"
      },
      "message": "Add markdown.frontmatter option to strip YAML front matter\n\nMarkdown files authored for static site generators (Jekyll, Hugo, etc.)\noften begin with a YAML front matter block delimited by \"---\" lines.\nGitiles has no front matter support, so CommonMark parses the block as\nordinary content: the closing \"---\" turns the metadata into a setext\n\u003ch2\u003e, rendering an ugly heading of raw \"key: value\" pairs at the top of\nthe page.\n\nAdd an opt-in \"markdown.frontmatter\" option (default false, consistent\nwith the other non-GitHub-flavor extensions). When enabled,\nGitilesMarkdown installs commonmark-ext-yaml-front-matter so the leading\nblock is parsed into a YamlFrontMatterBlock, and MarkdownToHtml omits\nthat block from the rendered output instead of throwing on the unknown\nCustomBlock.\n\nDocumented in Documentation/{config,markdown}.md and covered by\nGitilesMarkdownTest.\n\nChange-Id: I2ace7227728e964d2e73d5149a06f2d9a884f0c8\n"
    },
    {
      "commit": "e7402f5c7ea414b3d86df68e26aaf85ed2083908",
      "tree": "3c4d068c817206aa84de3bab9f5eb7e7456cd5f8",
      "parents": [
        "a1a207b53f52a567c6f27f52013303420c344189"
      ],
      "author": {
        "name": "Ajay Gupta",
        "email": "ajagup@qti.qualcomm.com",
        "time": "Fri Jul 03 10:45:17 2026 +0530"
      },
      "committer": {
        "name": "Ajay Gupta",
        "email": "ajagup@qti.qualcomm.com",
        "time": "Fri Jul 03 10:45:17 2026 +0530"
      },
      "message": "Avoid scanning every ref to render the commit log\n\nThe commit log page shows, next to each commit, the branches and tags\nthat point at it. Computing those labels loaded and peeled every\nreference in the repository on each page load. On repositories with very\nlarge reference counts, such as those backing a code review system where\nevery change adds several references, this work dominated the request\nand pushed memory use into the gigabytes and response times past a\nminute.\n\nOnly branches and tags are ever displayed, so the lookup now examines\njust those two namespaces rather than the whole set of references, which\nis far smaller and does not grow with review activity. A reference that\ncannot be resolved is skipped instead of failing the entire page,\npreserving the previous tolerance for damaged repositories.\n\nChange-Id: I4673f8821542b8ddb642257b20dc997c4e9ff6a7\n"
    },
    {
      "commit": "bcbf5b005af0c682b2e5f393324569dd2b3843ab",
      "tree": "b4d4006b5be370eaa1e8cd9842a3c945e46588a9",
      "parents": [
        "a1a207b53f52a567c6f27f52013303420c344189"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Jun 26 11:30:42 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Tue Jun 30 15:18:49 2026 +0200"
      },
      "message": "Bazel: Move Gitiles off JGit servlet-4 via EE8 bridge\n\nJGit master uses jakarta.servlet in servlet-facing modules while Gitiles\nstill runs on Jetty 12 EE8 and javax.servlet. Maintaining a Gitiles-only\nservlet-4 JGit branch is not sustainable; switch Gitiles to consume the\nJGit-owned EE8 servlet bridge so modules/jgit can advance to master.\n\nRewire //lib:jgit-servlet to export\n@jgit//org.eclipse.jgit.http.server.ee8:jgit-servlet-ee8. The generated\nEE8 jar keeps the original JGit Java packages and rewrites only servlet\nimports from jakarta.servlet to javax.servlet, so Gitiles code keeps its\nexisting imports. Gitiles already builds on Jetty EE8 and\njavax.servlet-api 4.0.1, so only the jgit-servlet dependency changes.\n\nBump the modules/jgit gitlink off the servlet-4 branch to the JGit\nmaster commit that provides the EE8 bridge targets (the same commit the\nGerrit companion pins). Re-pin external_deps.lock.json for the resulting\njgit-master dependency set.\n\nChange-Id: I9b9006e51f701c2c0db43178ae82e4691ec8eb68\n"
    },
    {
      "commit": "a1a207b53f52a567c6f27f52013303420c344189",
      "tree": "26d5ed5d6674b241955c514df8b0bda728206d95",
      "parents": [
        "d084706359ec66a6c775798633b924370b013e24"
      ],
      "author": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jun 24 09:41:13 2026 +0200"
      },
      "committer": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jun 24 09:41:13 2026 +0200"
      },
      "message": "Update JGit to c89601f\n\nand repin dependencies.\n\n$ git log --oneline --no-merges c46df0c..c89601f\n0c4bb4039 Update jna to 5.19.1\nb05364919 DfsObjDatabase: expose method to translate descriptions to PackList\n476c6a5c7 Checkout: Reduce temp filename truncation from 200 to 50 chars\n1c6d8987c Update jetty to 12.1.10\n60880a27d Update apache mina-sshd to 2.18.0\n5a995932e Update jna to 5.19.0\n4d9cc4caa Update bytebuddy to 1.18.10\n2972195ae Prepare 7.8.0-SNAPSHOT builds\n11b7c3500 Prepare 7.7.1-SNAPSHOT builds\ne15d6597f JGit v7.7.0.202606012155-r\na14a4be39 Log loose ref deletion during ref cleanup\n\nChange-Id: I689d14d980c7b282ec70c420be91143e26d2d9fc\n"
    },
    {
      "commit": "d084706359ec66a6c775798633b924370b013e24",
      "tree": "05cf925eb0b74629b6acd5d9b91d5d054944f1a7",
      "parents": [
        "e1ace85515ef8b6651a724e478c4bd2347e44fba"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Wed Jun 03 06:13:43 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Jun 12 07:40:41 2026 -0700"
      },
      "message": "Bump Jetty to 12.1.10 (preserve javax.servlet via jetty-ee8 adapter)\n\nFollow the same pattern as in the gerrit repository Id82de0bc91: upgrade\nthe embedded Jetty in the gitiles dev server to Jetty 12 while keeping\ngitiles on javax.servlet 4.0.1 via the jetty-ee8-* adapter modules.\n\nDependency changes mirror gerrit\u0027s:\n\n* tools/java_deps.MODULE.bazel: JETTY_VERSION 9.4.57.v20241219 -\u003e\n  12.1.10. Replace jetty-servlet with jetty-ee8-{nested,security,\n  servlet}. Add jetty-session (standalone module in Jetty 12).\n  Drop jetty-continuation (deprecated since Jetty 9, removed in\n  Jetty 10+).\n\n* lib/jetty/BUILD: expose new :nested target; :servlet exports\n  jetty-ee8-servlet + runtime_deps :nested; :security exports both\n  jetty-ee8-security and jetty-security; :server pulls in :session.\n\nDevServer.java migration to the Jetty 12 ee8 API:\n\n* org.eclipse.jetty.servlet.{ServletContextHandler,ServletHolder}\n  -\u003e org.eclipse.jetty.ee8.servlet.{...}.\n* The ee8 ServletContextHandler implements Supplier\u003ccore Handler\u003e;\n  appHandler() returns handler.get() so ContextHandlerCollection\n  receives the core Handler.\n* Jetty 12 replaced PathResource/URL-based ResourceHandler config\n  with ResourceFactory. setBaseResource takes a Resource produced\n  by ResourceFactory.root().newResource(Path). setDirectoriesListed\n  was renamed to setDirAllowed.\n\nA full Jakarta EE migration (jakarta.servlet, jetty-ee10 adapter)\nremains possible as separate follow-up work, matching gerrit\u0027s\ndeferral.\n\nBug: Issue 359909531\nChange-Id: I430d21d4191d69595f73041cd4ea7aa40b3c864d\n"
    },
    {
      "commit": "e1ace85515ef8b6651a724e478c4bd2347e44fba",
      "tree": "2cc15c597c6c72af178b0c54618f51b2d99ed60e",
      "parents": [
        "77214a27c8e599ad5683f037ebd85e2f1c9bd36c"
      ],
      "author": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Fri Jun 12 15:59:36 2026 +0200"
      },
      "committer": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Fri Jun 12 16:32:44 2026 +0200"
      },
      "message": "Update JGit to c46df0c50 and update MODULE.bazel.lock\n\nI854cc58ad6129a053fd98073ad04af99317bf8ce updated JGit to\neeba63ed4ebfe817628da5a221305b0c72af7b04 which doesn\u0027t exist neither\nin https://eclipse.gerrithub.io/eclipse-jgit/jgit\nnor in https://gerrit.googlesource.com/jgit.\n\nFix this by updating JGit to the current servlet-4 branch and update\nthe bazel lock file.\n\nIn addition add /modules/jgit/ to .bazelignore to prevent that\n`bazel test //...`\ntries to run jgit tests in the jgit submodule.\n\nChange-Id: Iaccbec4c901ee20e5e7d64186f5791b83dcd58c9\n"
    },
    {
      "commit": "77214a27c8e599ad5683f037ebd85e2f1c9bd36c",
      "tree": "73ba9f7c8a8ade2d04a97819b8e44693b393c164",
      "parents": [
        "4fdf6125211909c432887ea47e1e8ada37167953"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Sat May 30 07:41:23 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Jun 01 15:13:28 2026 +0200"
      },
      "message": "Bazel: Bump to 9.1.0\n\nUpgrade Bazel to 9.1.0 and refresh Bazel module dependencies to\nversions that are compatible with Bazel 9.\n\nThis requires a few migration adjustments:\n\n* pin aspect_bazel_lib 2.22.5 and rules_nodejs 6.7.4 at the root;\n  older versions (2.11.0 / 6.2.0) were resolved transitively via\n  the bazlets / aspect_rules_rollup deps, and their toolchain\n  definitions still pass the incompatible_use_toolchain_transition\n  keyword to rule(), which was removed in Bazel 9\n* bump rules_java 8.16.1 -\u003e 9.3.0 and rules_python 2.0.0-rc4 -\u003e 2.0.2\n* bump the java-prettify submodule to a revision that loads\n  java_library / java_binary from @rules_java//java:defs.bzl, since\n  rules_java 9 no longer injects them as native rules\n* regenerate MODULE.bazel.lock\n\nChange-Id: Ia72e44952d2f8aab09c85671ec44deffbf0369fd\n"
    },
    {
      "commit": "4fdf6125211909c432887ea47e1e8ada37167953",
      "tree": "641330b09ff37a11af115e5c1b4633b6e78cfdd3",
      "parents": [
        "60e7e35c7c8ee042d41b5889e7eb18dc7cead694"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri May 29 18:45:39 2026 +0200"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Mon Jun 01 15:13:15 2026 +0200"
      },
      "message": "Migrate to Bazel modules\n\nMirror the structure used in Gerrit: split MODULE.bazel into\ninclude() fragments under tools/ for bazlets, repos, and Java deps,\nand consume jgit as an in-tree bazel_dep via local_path_override.\n\nAll maven_jar declarations are replaced by a single RJE\nmaven.install in tools/java_deps.MODULE.bazel publishing the shared\n\"external_deps\" repository, consistent with Gerrit and JGit.\n\nThe modules/jgit submodule is bumped to a servlet-4 commit that\nships a proper MODULE.bazel so jgit can be consumed as a Bazel\nmodule.\n\nForce Gitiles\u0027 root-selected Maven versions for artifacts that may\nalso be contributed by JGit, keeping the shared Maven graph consistent\nunder Bzlmod.\n\nBazel version is bumped to 8.6.0 and rules_java is pinned to 8.16.1.\n\nJava version is updated to Java 21 by default, and the obsolete\nJava 17 Error Prone toolchain registration and definition are removed.\n\nBug: Issue 303819949\nChange-Id: I854cc58ad6129a053fd98073ad04af99317bf8ce\n"
    },
    {
      "commit": "60e7e35c7c8ee042d41b5889e7eb18dc7cead694",
      "tree": "98f4168c558ef0e848aca51ee0fbde0aa0d15bae",
      "parents": [
        "6d2033603424fcaad1a7b12e8abd3a67bde51758"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Wed Mar 18 12:04:15 2026 +0100"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri May 29 18:12:52 2026 +0200"
      },
      "message": "Update protobuf-java to 4.33.4 to match Gerrit\n\nChange-Id: Ifd8d7d3ef3d7cb3b114bd447a561629f738ac49d\n"
    },
    {
      "commit": "6d2033603424fcaad1a7b12e8abd3a67bde51758",
      "tree": "763ef0036b6e39d83a029f67e6ed738387f4db6c",
      "parents": [
        "44f4d076132452bd5ad79862062ec69b85690fb8"
      ],
      "author": {
        "name": "Sven Selberg",
        "email": "sven.selberg@axis.com",
        "time": "Mon May 25 01:33:52 2026 -0700"
      },
      "committer": {
        "name": "Sven Selberg",
        "email": "sven.selberg@axis.com",
        "time": "Mon May 25 01:34:57 2026 -0700"
      },
      "message": "docs: Correct header level for log/limiting parameters\n\nChange-Id: Ibfeb4736ab5379fed38ca6f6799e3d8bd8bfbea4\n"
    },
    {
      "commit": "44f4d076132452bd5ad79862062ec69b85690fb8",
      "tree": "12d11c5f5346863271725ac113eff264e3714760",
      "parents": [
        "c4d9bd990a00bf097239409e618a61327c8776d1"
      ],
      "author": {
        "name": "Sven Selberg",
        "email": "svense@axis.com",
        "time": "Fri May 22 08:58:56 2026 +0200"
      },
      "committer": {
        "name": "Sven Selberg",
        "email": "svense@axis.com",
        "time": "Fri May 22 08:58:56 2026 +0200"
      },
      "message": "chore: Fix typo INCORECT_PARAMETER -\u003e INCORRECT_PARAMETER\n\nChange-Id: Ib740dd03d225befb9735b248c28c9f92bd012bd9\n"
    },
    {
      "commit": "c4d9bd990a00bf097239409e618a61327c8776d1",
      "tree": "9a1100448d8518b030373be94b1f3bc0872dcdb5",
      "parents": [
        "7a34d138a61916f3c0e1ed540b2a4f7c8eb35375"
      ],
      "author": {
        "name": "Sven Selberg",
        "email": "sven.selberg@axis.com",
        "time": "Mon May 18 23:15:58 2026 -0700"
      },
      "committer": {
        "name": "Sven Selberg",
        "email": "svense@axis.com",
        "time": "Tue May 19 09:47:06 2026 +0200"
      },
      "message": "docs: Document author and committer log parameters\n\nFormat the +log api documentation to make it more readable.\n\nIssue: 376381593\nChange-Id: Ia96889d6a16fc88eb27758b4e4915890cf6e9adb\n"
    },
    {
      "commit": "7a34d138a61916f3c0e1ed540b2a4f7c8eb35375",
      "tree": "3df1163bb7d4e1e66703ce61fa3202e30a1489ff",
      "parents": [
        "9df3a168e43bcb41015c16873c095ee7f9907542"
      ],
      "author": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Fri May 15 00:46:48 2026 +0530"
      },
      "committer": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Tue May 19 13:02:33 2026 +0530"
      },
      "message": "LogServlet: support grep for log messages\n\nAdd a `grep` query parameter to the log servlet. This lets Gitiles filter\nlog results by commit message using JGit\u0027s message filter.\n\nThe new filter is combined with the existing log filters, so it can be\nused together with author or committer filtering.\n\nAdd coverage that log grep stays in the log page navigation URD.\n\nIssue: 376381593\nChange-Id: Ib1ad5fe6cfdb29ce3528aedad983126d1f7590ed\n"
    },
    {
      "commit": "9df3a168e43bcb41015c16873c095ee7f9907542",
      "tree": "daaca0bec75bd1a7f8eec38f97dcfa902fe78bfe",
      "parents": [
        "1ffd4a1f72a4ba13f2e3a3c544c1428caaa2e330"
      ],
      "author": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Thu May 14 00:02:15 2026 +0530"
      },
      "committer": {
        "name": "Devansh Varshney",
        "email": "devansh@libreoffice.org",
        "time": "Thu May 14 00:02:15 2026 +0530"
      },
      "message": "LogServletTest: cover author and committer log filters\n\nGitiles already supports filtering the log by author and committer\nthrough query parameters.\n\nAdd servlet coverage for this existing behavior before building more\nlog search support on top of it.\n\nIssue: 376381593\nChange-Id: I83aec3e5066c0591e49641228d69c285401382a4\n"
    },
    {
      "commit": "1ffd4a1f72a4ba13f2e3a3c544c1428caaa2e330",
      "tree": "afae0143659f5371cb3f2e00b3f590d3401dfbc0",
      "parents": [
        "d113c2721180bc0fe0da07950a4326e7bc4f5c3a"
      ],
      "author": {
        "name": "Jackson Toeniskoetter",
        "email": "jackdt@google.com",
        "time": "Tue Mar 31 14:40:08 2026 -0700"
      },
      "committer": {
        "name": "Jackson Toeniskoetter",
        "email": "jackdt@google.com",
        "time": "Wed Apr 01 15:37:38 2026 -0700"
      },
      "message": "In ViewFilter, if a revision cannot be found then throw an exception\ninstead of returning null.\n\nChange-Id: I9c00409c67688d319f6c440afa4ad73fb5ecb7a9\n"
    },
    {
      "commit": "d113c2721180bc0fe0da07950a4326e7bc4f5c3a",
      "tree": "4c05a3793819e47e4f109e29c8b75cd4d98f67b2",
      "parents": [
        "585d37dae3dab933127c7f1d67acb49d7181bb77",
        "b560a15ef9113934d66377e16de8beb95cbeb0b4"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Mon Feb 23 16:20:19 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Mon Feb 23 16:20:19 2026 -0800"
      },
      "message": "Merge \"Add missing commons-io dependency in BUILD file\""
    },
    {
      "commit": "b560a15ef9113934d66377e16de8beb95cbeb0b4",
      "tree": "d1eac13d0b5892e52e83c94a100ea7306f776b90",
      "parents": [
        "e8b6a00f55823e5a328e4ba2c7e51f048188a292"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 20 19:02:42 2026 -0800"
      },
      "committer": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Mon Feb 23 16:05:57 2026 -0800"
      },
      "message": "Add missing commons-io dependency in BUILD file\n\nAdding //lib:commons-io to DEPS in java/com/google/gitiles/BUILD ensures\nthat archive generation (used by ArchiveServlet) has the required\nCountingOutputStream class available at runtime.\n\nThis fixes the NoClassDefFoundError occurring when using tgz links.\n\nChange-Id: Id7551f10385e01f6f9db6a73d1f5f877fd55c281\n"
    },
    {
      "commit": "585d37dae3dab933127c7f1d67acb49d7181bb77",
      "tree": "6eb016f7a7bf6ab74041dcc0fe522a26d30243c8",
      "parents": [
        "e8b6a00f55823e5a328e4ba2c7e51f048188a292"
      ],
      "author": {
        "name": "Nitzan Gur-Furman",
        "email": "nitzan@google.com",
        "time": "Sun Feb 22 23:32:21 2026 -0800"
      },
      "committer": {
        "name": "Nitzan Gur-Furman",
        "email": "nitzan@google.com",
        "time": "Sun Feb 22 23:32:21 2026 -0800"
      },
      "message": "Add @RunWith(JUnit4.class) to ArchiveServletTest.\n\nChange-Id: Ia2440b53ffe9cee645246dbc04ee31e64f1f2190"
    },
    {
      "commit": "e8b6a00f55823e5a328e4ba2c7e51f048188a292",
      "tree": "e96cadcac94a2c83a93c5ec7e4c58033b2556cef",
      "parents": [
        "080b31ca6b34b044d92697cad9b696c1cf43f1d8"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 20 18:33:41 2026 -0800"
      },
      "committer": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 20 18:51:38 2026 -0800"
      },
      "message": "Fix 080b31c for updated protobuf dep and alpha ordering\n\n080b31c updated Gitiles dependencies but got merged too early\n\nThis change does the following:\n* Reports that diffutils moved from\ncom.googlecode.java-diff-utils:diffutils to\nio.github.java-diff-utils:java-diff-utils\n* Updates protobuf from 3.25.5 to 4.33.2\n* Alphabetizes the new common-io dependency\n\nChange-Id: I79c5ea1ee2081eba77cbfce9cd7762709666453e\n"
    },
    {
      "commit": "080b31ca6b34b044d92697cad9b696c1cf43f1d8",
      "tree": "1800c0e7bf718a7e6e10fe7b754e6521d6111e43",
      "parents": [
        "279e12103575b1c08223a8feacb1e62349a9b71f"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 13 16:28:39 2026 -0800"
      },
      "committer": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 20 10:54:47 2026 -0800"
      },
      "message": "Update Gitiles dependencies\n\nThis area was overdue for attention, there are many bugs filed\nabout out-of-date Gitiles dependencies.\n\nGoogle uses its internal build system to build Gitiles, and we\nexperience friction in updating internally, since we need to update\nnot just Gitiles, but all internal uses of that dependency. With\nAI assistance I updated WORKSPACE library versions to match the\nones Google is using, if Google is using a newer version.\n\nAll of these updates involve stable APIs, no source file updates\nwere required.\n\nUpdating commons-compress added a new dependency on commons-io,\nso added that (thanks to David for finding that, Diffy for the\nwin!)\n\nTesting was to run `bazel test //...`, run the `tools/run.dev`\nscript and test localhost, and run `tools/maven/mvn.sh install`.\n\nI compared these updated versions with the versions in the JGit\nand Gerrit projects. All permutations are now there--Gitiles uses\na more recent version than the other two, JGit does, and Gerrit\ndoes. I can run the same update on JGit and Gerrit once this\nchange is in.\n\n+------------------------+-------------------+--------------------+\n| Library                | Old Version       | New Version        |\n+------------------------+-------------------+--------------------+\n| ASM (OW2)              | 9.2               | 9.9.1              |\n| Autolink               | 0.10.0            | 0.11.0             |\n| Byte Buddy             | 1.14.9            | 1.18.2             |\n| Commonmark             | 0.21.0            | 0.24.0             |\n| Commons Codec          | 1.16.0            | 1.18.0             |\n| Commons Compress       | 1.25.0            | 1.28.0             |\n| Commons IO             | \u003cnone\u003e            | 2.21.0             |\n| Commons Lang3          | 3.8.1             | 3.18.0             |\n| Commons Text           | 1.2               | 1.10.0             |\n| Diffutils              | 1.3.0             | 4.16               |\n| Error Prone Annot.     | 2.22.0            | 2.46.0             |\n| Gson                   | 2.10.1            | 2.13.2             |\n| Guava                  | 32.1.2-jre        | 33.5.0-jre         |\n| Guava Failureaccess    | 1.0.1             | 1.0.3              |\n| Hamcrest               | 2.2               | 3.0                |\n| ICU4J                  | 74.2              | 78.2               |\n| Jetty                  | 9.4.53.v20231009  | 9.4.57.v20241219   |\n| Junit                  | 4.12              | 4.13.2             |\n| Mockito                | 5.6.0             | 5.21.0             |\n| Objenesis              | 2.6               | 3.0.1              |\n| Protobuf               | 3.19.4            | 3.25.5             |\n| SLF4J                  | 1.7.36            | 2.0.17             |\n| Servlet API            | 3.1.0             | 4.0.1              |\n| Truth                  | 1.1               | 1.4.4              |\n| XZ                     | 1.9               | 1.11               |\n+------------------------+-------------------+--------------------+\n\nRelease-Notes: Stabilize WORKSPACE dependencies and fix checksums.\nChange-Id: Ic7dc1c8e96e8c1b26c80d259fb0b194ae4fc41f8\n"
    },
    {
      "commit": "279e12103575b1c08223a8feacb1e62349a9b71f",
      "tree": "c7ea6368ef04e04fed504b3900ad86a7fb86c8f7",
      "parents": [
        "f8b815b8b8e74136d0c1b45ae90bd4f66ba5916f",
        "6a98ba594edd03b030e70efa50097c00ea81d202"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 20 08:00:30 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Fri Feb 20 08:00:30 2026 -0800"
      },
      "message": "Merge \"Update Gitiles bazel version to 7.6.1 (matches JGit \u0026 Gerrit)\""
    },
    {
      "commit": "f8b815b8b8e74136d0c1b45ae90bd4f66ba5916f",
      "tree": "07121dd413ae2c99e870e4af5dc6756864f235bb",
      "parents": [
        "9b6c331d6f24db2d0faae016f801effb39a2de2f"
      ],
      "author": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Feb 20 07:50:54 2026 +0100"
      },
      "committer": {
        "name": "David Ostrovsky",
        "email": "david@ostrovsky.org",
        "time": "Fri Feb 20 08:08:46 2026 +0100"
      },
      "message": "Add archive servlet integration test\n\nIn change Ic7dc1c8e96e an update of commons-compress from 1.25.0 to\n1.28.0 was attempted as part of a broader WORKSPACE dependency refresh.\n\ncommons-compress 1.28.0 introduces a runtime dependency on\ncommons-io (via org.apache.commons.io.output.CountingOutputStream),\nwhich is used by JGit\u0027s archive implementation\n(org.eclipse.jgit.archive.TarFormat / TgzFormat) when writing\ntar/tgz archives.\n\nGitiles\u0027 Bazel configuration does not automatically pull in Maven\ntransitive dependencies, so commons-io must be declared explicitly.\nThe attempted upgrade initially omitted this dependency, which would\nhave caused runtime failures when serving archive downloads, e.g.:\n\n  /\u003crepo\u003e/+archive/refs/heads/\u003cref\u003e.tar.gz\n\nresulting in:\n\n  java.lang.NoClassDefFoundError:\n    org/apache/commons/io/output/CountingOutputStream\n\nThis issue was identified during code review and corrected before\nsubmission, but the existing test suite did not exercise the archive\nservlet code path and therefore did not detect the breakage.\n\nAdd a servlet-level integration test for the archive endpoint that:\n\n* creates a small in-memory test repository\n* requests a .tar.gz archive via GitilesServlet\n* verifies HTTP 200 response\n* verifies response body is non-empty\n\nThis test executes ArchiveCommand.call() end-to-end and will fail\nwith NoClassDefFoundError if commons-io is missing from the runtime\nclasspath, preventing similar dependency regressions in future\ndependency updates.\n\nChange-Id: Ia9266aa6a6ca63b119e2216d8376ce0382ae6e44\n"
    },
    {
      "commit": "6a98ba594edd03b030e70efa50097c00ea81d202",
      "tree": "9e1838c066b45fe5afd06ac3988c8d858b415a59",
      "parents": [
        "9b6c331d6f24db2d0faae016f801effb39a2de2f"
      ],
      "author": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 13 15:36:44 2026 -0800"
      },
      "committer": {
        "name": "Terry Parker",
        "email": "tparker@google.com",
        "time": "Fri Feb 13 15:42:35 2026 -0800"
      },
      "message": "Update Gitiles bazel version to 7.6.1 (matches JGit \u0026 Gerrit)\n\nAlso fix bazel\u0027s jdk tool registrations to match JGit \u0026 Gerrit.\n\nChange-Id: Iec03b7e28a8d20f49f27eb399c855cf4ffda7245\n"
    },
    {
      "commit": "9b6c331d6f24db2d0faae016f801effb39a2de2f",
      "tree": "1bdfbd91b35b9b17b69498a85c239fa2a5ca47fd",
      "parents": [
        "dd5fd429d1362bfc119c8bbd56235ff79c5662fd",
        "7fc3f09cab178489f80629a0c7940bb217fe7dc2"
      ],
      "author": {
        "name": "Theodore Dubois",
        "email": "tbodt@google.com",
        "time": "Wed Jan 28 14:49:35 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Jan 28 14:49:35 2026 -0800"
      },
      "message": "Merge \"Make copy/pasting breadcrumbs produce valid file paths\""
    },
    {
      "commit": "7fc3f09cab178489f80629a0c7940bb217fe7dc2",
      "tree": "62a14d36153553e3888574e373568aba003d10a0",
      "parents": [
        "ff3f30bb02ae529910784caffa2499596af49997"
      ],
      "author": {
        "name": "Theodore Dubois",
        "email": "tbodt@google.com",
        "time": "Wed Jan 28 14:26:17 2026 -0800"
      },
      "committer": {
        "name": "Theodore Dubois",
        "email": "tbodt@google.com",
        "time": "Wed Jan 28 14:41:51 2026 -0800"
      },
      "message": "Make copy/pasting breadcrumbs produce valid file paths\n\nBug: 479282458\nChange-Id: Ice76ba569b6f2b224676a5ad861b10746a6a6964\n"
    },
    {
      "commit": "dd5fd429d1362bfc119c8bbd56235ff79c5662fd",
      "tree": "000b6204ff002884b024ab27b936172119f4d819",
      "parents": [
        "ff3f30bb02ae529910784caffa2499596af49997",
        "8c3d87d6123823de8e4f515538aa93ef37210cb1"
      ],
      "author": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@gmail.com",
        "time": "Wed Jan 28 13:34:47 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Wed Jan 28 13:34:47 2026 -0800"
      },
      "message": "Merge changes I0f844eaa,Ife1777c0\n\n* changes:\n  Fix dev server configuration\n  Format source code using google-java-format 1.24.0\n"
    },
    {
      "commit": "8c3d87d6123823de8e4f515538aa93ef37210cb1",
      "tree": "5530e2be9bee831ddc4f6d288c9fbe6e0188e4ac",
      "parents": [
        "318b4a1d0f29f7a1e1dd5d25f242c12e2f4f1e6f"
      ],
      "author": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jan 28 11:09:47 2026 +0100"
      },
      "committer": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jan 28 11:24:58 2026 +0100"
      },
      "message": "Fix dev server configuration\n\nThe config option gitiles.basePath was overloaded in [1]\nwhich broke the default behavior of gitiles in dev mode. It used to scan\nthe path configured via gitiles.basePath to find git repositories.\n\nWith [1] it now sets the servlet context path for gitiles to the value\nof the same option which doesn\u0027t work since the dev mode by default sets\ngitiles.basePath to the current working directory which doesn\u0027t work\nas a servlet context path which results in gitiles returning\n\"404 Not Found\" instead of showing the list of repositories found under\nbasePath.\n\nFix this by renaming the option introduced in [1] to\ngitiles.contextPath and adapt the documentation accordingly.\n\n[1] I1da95745d1bc8eaa7d5b204a942bcee0b6a027a8\n\nChange-Id: I0f844eaaaadbfd21161e04b915b541a534bbb4a6\n"
    },
    {
      "commit": "318b4a1d0f29f7a1e1dd5d25f242c12e2f4f1e6f",
      "tree": "81ad2c0152f4f02227e7ab4a503390ae7686d7b0",
      "parents": [
        "fbbb2969d93aef88173e0f0f59b77bd8b9f3d49d"
      ],
      "author": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jan 28 11:19:27 2026 +0100"
      },
      "committer": {
        "name": "Matthias Sohn",
        "email": "matthias.sohn@sap.com",
        "time": "Wed Jan 28 11:19:27 2026 +0100"
      },
      "message": "Format source code using google-java-format 1.24.0\n\nFollow gerrit which updated to google-java-format 1.24.0 in\nI8607721e9e83519fa2bab415462ace5255032068.\n\nChange-Id: Ife1777c0fea0cd002422960aed481bc46a955fb0\n"
    },
    {
      "commit": "ff3f30bb02ae529910784caffa2499596af49997",
      "tree": "a15f3bcda4eea72477246b31c35801204a29847e",
      "parents": [
        "91f481a71e22fe1438b6acea17f01f5e2422b965",
        "1b8d8e325823ed13b1ac97c8b68a28c0aac121b5"
      ],
      "author": {
        "name": "Adrià Vilanova Martínez",
        "email": "me@avm99963.com",
        "time": "Tue Jan 27 15:22:14 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Jan 27 15:22:14 2026 -0800"
      },
      "message": "Merge \"Make metadata tables responsive\""
    },
    {
      "commit": "91f481a71e22fe1438b6acea17f01f5e2422b965",
      "tree": "d9f77226bcc4f87e5f046f5d5198c9c92dcff67c",
      "parents": [
        "2b3e80a07b5bd873b00610cb9113544e22dedcb8",
        "bff35388f5507d076b401592f2ce195ee04f6c9f"
      ],
      "author": {
        "name": "Xing Huang",
        "email": "xingkhuang@google.com",
        "time": "Tue Jan 27 08:32:20 2026 -0800"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Jan 27 08:32:20 2026 -0800"
      },
      "message": "Merge \"LogServlet: Handle HEAD symref to non-existent branch\""
    },
    {
      "commit": "2b3e80a07b5bd873b00610cb9113544e22dedcb8",
      "tree": "c95062bd5910d2080430ce82511d43bbfadbe16b",
      "parents": [
        "fbbb2969d93aef88173e0f0f59b77bd8b9f3d49d"
      ],
      "author": {
        "name": "Adrià Vilanova Martínez",
        "email": "me@avm99963.com",
        "time": "Thu Jan 15 22:11:16 2026 +0100"
      },
      "committer": {
        "name": "Adrià Vilanova Martínez",
        "email": "me@avm99963.com",
        "time": "Tue Jan 27 05:02:53 2026 -0800"
      },
      "message": "Clarify development server URL\n\nSince commit b1b8c4a45d5feb5bd11798a7d8469f7af0f04405, the dev server\nuses by default a base path equal to the absolute path of the current\ndirectory.\n\nThis CL clarifies this in the developer guide.\n\nChange-Id: I3c609736008760fad5748244960d82ca7595a590\n"
    },
    {
      "commit": "1b8d8e325823ed13b1ac97c8b68a28c0aac121b5",
      "tree": "31fadd1557b8608087b31452f5ad9bac8215f44c",
      "parents": [
        "fbbb2969d93aef88173e0f0f59b77bd8b9f3d49d"
      ],
      "author": {
        "name": "Adrià Vilanova Martínez",
        "email": "me@avm99963.com",
        "time": "Mon Dec 27 23:58:01 2021 +0100"
      },
      "committer": {
        "name": "Adrià Vilanova Martínez",
        "email": "me@avm99963.com",
        "time": "Tue Jan 27 04:39:22 2026 -0800"
      },
      "message": "Make metadata tables responsive\n\nIn commit 9b8a1eb9ea3a6945c82411ef6742637f11b6056f we made Gitiles\nresponsive so that it looked better in small screens. However, metadata\ntables weren\u0027t made responsive due to some accessibility challenges.\n\nThis CL finishes the responsiveness refactor by also making metadata\ntables responsive. We achieve this by changing the HTML table with a\ndescription list styled with a grid layout and subgrids[1] that adapt to\nthe screen width.\n\nWe use description lists since they are more semantically accurate for\nkey-value metadata pairs than tables.[2]\n\nThe linked bug contains screenshots that show the change.\n\n[1]: Subgrids achieved the Baseline status in 2023:\n     https://web.dev/articles/css-subgrid.\n[2]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-dl-element\n\nBug: https://github.com/google/gitiles/issues/235\nChange-Id: I83026d0fbe7df7601a07be5e6122cdbb2d27ec81\n"
    },
    {
      "commit": "bff35388f5507d076b401592f2ce195ee04f6c9f",
      "tree": "acb9de58e85d3c5b9880c3c20c1427abac50fd2c",
      "parents": [
        "fbbb2969d93aef88173e0f0f59b77bd8b9f3d49d"
      ],
      "author": {
        "name": "Xing Huang",
        "email": "xingkhuang@google.com",
        "time": "Mon Jan 26 15:08:03 2026 -0600"
      },
      "committer": {
        "name": "Xing Huang",
        "email": "xingkhuang@google.com",
        "time": "Mon Jan 26 15:43:39 2026 -0600"
      },
      "message": "LogServlet: Handle HEAD symref to non-existent branch\n\nThe ObjectId of a Ref can be null. If HEAD is a symbolic ref to a\nnon-existent branch, headRef.getObjectId() will return null. The\ncurrent LogServlet code does not handle this, causing\nNullPointerException in RevWalk.\n\nThis issue is particularly common when a repo is created but nothing is\npushed yet; so a HEAD symmetry is created to point to refs/heads/main by default but lacks a commit SHA.\n\nTreat a HEAD ref with a null ObjectId as if the ref itself were null,\nreturning no view instead of throwing an internal error.\n\nChange-Id: I4fe85010db0694403770923fcdf957854e72cd34\nSigned-off-by: Xing Huang \u003cxingkhuang@google.com\u003e\n"
    },
    {
      "commit": "fbbb2969d93aef88173e0f0f59b77bd8b9f3d49d",
      "tree": "af445b18475c09ccf0fbb49ad0e3127410de40c3",
      "parents": [
        "97d418e1895740a2ee10eb54e6279ee762aaaa35"
      ],
      "author": {
        "name": "liyang",
        "email": "leeyoung.li@outlook.com",
        "time": "Tue Oct 07 21:53:01 2025 +0800"
      },
      "committer": {
        "name": "liyang",
        "email": "leeyoung.li@outlook.com",
        "time": "Wed Oct 15 12:23:26 2025 +0800"
      },
      "message": "Fix package WAR missing failureaccess JAR\n\nwebserver ClassNotFoundException due to WAR missing failureaccess JAR\nThe pkg_war() rule is missing this runtime dependency. Without it we get\nexceptions like:\njava.lang.ClassNotFoundException:\ncom.google.common.util.concurrent.internal.InternalFutureFailureAccess\n\nBug: Issue 450323474\nChange-Id: Icfb1f0e7625b5e64a01ca1552bc579b202ffb83a\n"
    },
    {
      "commit": "97d418e1895740a2ee10eb54e6279ee762aaaa35",
      "tree": "c9b12547365b9f6e78ea328d6d3461288393a6ba",
      "parents": [
        "b89ae948646625812b308b902a617ea63317a880"
      ],
      "author": {
        "name": "Antoine Musso",
        "email": "hashar@free.fr",
        "time": "Tue Sep 30 19:35:40 2025 +0200"
      },
      "committer": {
        "name": "Antoine Musso",
        "email": "hashar@free.fr",
        "time": "Tue Sep 30 19:39:56 2025 +0200"
      },
      "message": "Fix edit link when base URL has a path\n\nOn a setup having a path prefix such as `/r` the edit link woud be\nbroken because the prefix is stripped.\n\nThe reason is `URI.resolve()` is given an absolute path which strip the\npath from the base URI:\n\n  jshell\u003e new URI(\"https://example.org/r\").resolve(\"/admin/repos/edit\");\n                                      ^^\n  $2 \u003d\u003d\u003e https://example.org/admin/repos/edit\n                           ^^^____oops\n\nWhich is consistent with python `os.path`:\n\n  \u003e\u003e\u003e os.path.join(\u0027https://example.org/r\u0027, \u0027/admin/repos/edit\u0027);\n  \u0027/admin/repos/edit\u0027\n\nRemove the leading slash from the edit URI and update the test with the\nproper expected value.\n\nBug: https://phabricator.wikimedia.org/T405481\nChange-Id: I98e9d89cbc4850270f81b52665f2c8849f25c879\n"
    },
    {
      "commit": "b89ae948646625812b308b902a617ea63317a880",
      "tree": "5fa02b6b07c98990dfa8815cac0daeb79f3efa4f",
      "parents": [
        "c0094a91f4022f397396c486b5fdc6f814ae51af"
      ],
      "author": {
        "name": "Antoine Musso",
        "email": "hashar@free.fr",
        "time": "Tue Sep 30 19:31:35 2025 +0200"
      },
      "committer": {
        "name": "Antoine Musso",
        "email": "hashar@free.fr",
        "time": "Tue Sep 30 19:39:52 2025 +0200"
      },
      "message": "Tests: update path prefix in git urls\n\nWhen testing URLs, it is helpfull to distinguish between the base URL\n(which can contains a path prefix such as `/r`) and the application\npaths that is appended to it.\n\nUsing `/foo` is a bit meaningless, replace it by `/path-prefix` which is\nself describing and helps differentiating between the base URL and the\nrest of the path.\n\nChange-Id: I91230750cc67b7a2876e4da7779a0b0e20f62639\n"
    },
    {
      "commit": "c0094a91f4022f397396c486b5fdc6f814ae51af",
      "tree": "5c56efe245a2bbdf379d769c019e4157bf4e035a",
      "parents": [
        "0a3493eb4932ed3df16d253cbbb9f79709ace43c",
        "b1b8c4a45d5feb5bd11798a7d8469f7af0f04405"
      ],
      "author": {
        "name": "Adithya Chakilam",
        "email": "achakilam@nvidia.com",
        "time": "Tue Sep 09 09:13:41 2025 -0700"
      },
      "committer": {
        "name": "Gerrit Code Review",
        "email": "noreply-gerritcodereview@google.com",
        "time": "Tue Sep 09 09:13:41 2025 -0700"
      },
      "message": "Merge \"Add ability to set base path in dev mode\""
    }
  ],
  "next": "0a3493eb4932ed3df16d253cbbb9f79709ace43c"
}
