Re: [RFC] test: build python docs

[David Bremner <david@tethera.net>]

David Bremner <david@tethera.net> writes:

> The idea is to consider any warning from sphinx-build as a test
> failure. This requires the -W option, which we probably don't want for
> builds.
> ---
>
> I'm not sure the best way to do this. An alternative approach would be
> to build the python docs during the main build (much as we currently
> build the ruby bindings). I suspect we'd want some way to disable the
> "-W" option to sphinx-build in that case, so that python doc failures
> don't break the entire build. On the other hand, for distros breaking
> the test suite is pretty much equivalent to breaking the build. In the
> proposed patch the test could be disabled by dropping sphinx-build
> from the test env (which makes the man pages go away), or by patching
> the single test out.

Sure enough, no sooner do I send this and fix the warning about
get_config_list, I get others that I don't see an obvious fix for:

/home/bremner/software/upstream/notmuch/bindings/python/docs/source/status_and_errors.rst:17: WARNING: error while formatting arguments for notmuch.STATUS: 'Status' object has no attribute '__bases__'
/home/bremner/software/upstream/notmuch/bindings/python/docs/source/threads.rst:14: WARNING: don't know which module to import for autodocumenting u'__str__' (try placing a "module" or "currentmodule" directive in the document, or giving an explicit module name)