docs/concept/troubleshooting.soy

{namespace buck.troubleshooting}

/***/
{template .soyweb}
  {call buck.page}
    {param title: 'Troubleshooting' /}
    {param content}

If Buck stops working, then there are several things that you can try to do to
fix it.


<h2>Make sure you are using the Oracle JDK</h2>

Buck has only been tested with the Oracle JDK. If you are using an alternative,
such as OpenJDK, then things are not guaranteed to work.


<h2>Rebuild Buck</h2>

In Buck's short history, its autoupdate logic has had several bugs.
If you think that Buck failed to autoupdate correctly (or failed to rebuild
correctly after an autoupdate, which is likely if you killed Buck in the
middle of an autoupdate), then your best bet is to rebuild it yourself:

{literal}<pre>
cd &lt;directory-where-you-checked-out-Buck>
git checkout master
git pull --rebase
ant clean jar
</pre>{/literal}


<h2>Run `buck clean`</h2>

Ideally, this solution will never work. Seriously.
If Buck is working correctly, then it should know which files have been
modified and which files need to be rebuilt.
<p>
That said, Buck is not perfect, so it is possible that you have found a defect.
In this case, give <code>buck clean</code> a shot and file a bug if you have
found a reproducible bug.


<h2>Make sure you do not have a .nobuckcheck file</h2>

If you have been working on Buck itself, you had to create
a <a href="nobuckcheck.html"><code>.nobuckcheck</code></a> file to disable
Buck's autoupdating capability
(assuming you are working on a project that uses
{sp}<a href="buckversion.html"><code>.buckversion</code></a>).
Once you are done modifying Buck, you should delete the
{sp}<code>.nobuckcheck</code> file so that Buck autoupdates again.
<p>
If you forget to do this, it is inevitable that someone will update your
project such that it requires a newer version of Buck, and your local build
will inexplicably stop working because your are using an outdated version of
Buck. The alternative is to periodically <code>git pull --rebase</code> to
update your checkout of Buck manually, but that is error-prone.


<h2>Delete all generated files in your project.</h2>

Buck is designed so that all generated files are written to the
{sp}<code>buck-out</code> directory, which makes <code>buck clean</code> trivial
to implement. However, you may use additional tools (such as an IDE) that
generate files in other parts of the tree. Such files may inadvertently get
included via <a href="{ROOT}function/glob.html"><code>glob()</code></a> rules,
which would interfere with Buck.
<p>
For example, if you are using Git, then you can run:

<pre>git clean -xfdn</pre>

to get a list of files in your project that are
not under version control. The <code>-n</code> switch is for "dry run,"
which means that Git will not delete any files when you run
{sp}<code>git clean</code>. If you want to use Git to remove the generated
files while preserving some non-versioned files (such
as <code>local.properties</code>), then use it with the <code>-e</code> switch:

<pre>git clean -xfd -e local.properties</pre>

Note that <code>-e</code> can be specified multiple times.

    {/param}
  {/call}
{/template}