Conversation
Without this, installed packages don't work very well for indexing
|
The idea is sane, though we should probably use MANIFEST.in instead. I vaguely recall package_data being a distutils thing, obsolete, and unreliable in some way. I can see us leaving trilite out of this mix for now, since we have not-quite-horrible instructions in deployment.mkd about installing it. However, perhaps it would be friendlier to install it, too. |
…nstall" installations can actually find their templates. We have yet to figure out how to ship compiled plugins, like the clang one. Ref mozilla#110.
|
We could probably make a custom distutils Then doing |
|
Moved the automatically-run-make thread to a dedicated bug: https://bugzilla.mozilla.org/show_bug.cgi?id=881490. |
Without this, installed packages don't work very well for indexing.
Note that I don't know my way around setuptools/distribute/whatever so don't trust me on this - it's more of a bug report that not doing this doesn't work, I guess? Anyway, with
setup.py install, it didn't include the .so so actually using clang wasn't very productive. If we're going to ship the rest of the clang plugin anyway...