Closed Bug 761105 Opened 12 years ago Closed 12 years ago

egg-info directories checked in the source tree

Categories

(Firefox Build System :: General, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: glandium, Assigned: glandium)

Details

Attachments

(1 file)

Thanks to bug 661908, egg-info directories are created in the source tree (bug 758823). But there's something even better: there are egg-info directories checked in mercurial:
- build/pylib/blessings/blessings.egg-info
- other-licenses/virtualenv/virtualenv.egg-info

The latter comes from bug 661908 itself, but the former pre-dates it. So running configure may modify files under build/pylib/blessings/blessings.egg-info (and it happened to me in https://hg.mozilla.org/try/rev/7c9f4fa9c5a1 )
I just downloaded both of these packages from PyPI and both tar.gz archives contain egg-info directories.

I thought egg-info directories from source packages provided (possibly important) metadata. What would removing them accomplish?

As far as the blessings metadata diff, it is listed as version 1.0 in the upstream package. When I checked in the virtualenv integration a few days ago, I checked in the bump to 1.1. However, it appears it was regressed (by accident, likely) in 0d736d4fa057. *sigh*

To further complicate matters, my work in bug 751795 may require further changes to how the Python environment is configured. For that to be a true front-end to the build system, it needs to wrap configure, which is where virtualenv is being created currently. Ugh. But, that's for a different time, luckily.

Why does Python packaging have to be so complex?
Beats me, man.
(In reply to Gregory Szorc [:gps] from comment #2)
> I just downloaded both of these packages from PyPI and both tar.gz archives
> contain egg-info directories.
> 
> I thought egg-info directories from source packages provided (possibly
> important) metadata. What would removing them accomplish?

Avoiding this:
> When I checked in the virtualenv integration a few days
> ago, I checked in the bump to 1.1. However, it appears it was regressed (by
> accident, likely) in 0d736d4fa057. *sigh*


> Why does Python packaging have to be so complex?

The alternative question is: do we really have to run setup.py on these things when we never had to? It might be easier for virtualenv, but it really feels we'd have much less problems if we "installed" that stuff in the virtualenv "manually".
Attachment #629736 - Flags: review?(ted.mielczarek) → review+
https://hg.mozilla.org/mozilla-central/rev/77f792cd4483
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
Product: Core → Firefox Build System
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: