Support .egg installations - [merged]

[jaraco]

In GitLab by @asottile on May 11, 2019, 22:01

Merges egg_support -> master

Signed-off-by: Anthony Sottile [email protected]

[jaraco]

In GitLab by @asottile on May 11, 2019, 22:02

I don't want to support eggs either, but it's a blocker for using importlib-metadata for pytest via pluggy

See pytest-dev/pluggy#205 for more details

[jaraco]

In GitLab by @codecov on May 11, 2019, 22:06

Codecov Report

Merging #65 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master    #65   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files           4      4           
  Lines         202    202           
  Branches       19     19           
=====================================
  Hits          202    202

Impacted Files	Coverage Δ
importlib_metadata/_hooks.py	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update eb107b0...ef0ce71. Read the comment docs.

[jaraco]

In GitLab by @jaraco on May 12, 2019, 12:27

Reading the downstream issue, here's the summary:

When importlib-metadata is installed as an egg, it's unusable because it can't read its own metadata.

I'd say it's far more of a blocker that the package is unusable as an egg than for it to support eggs.

However, if adding support for eggs addresses the blocker and is likely to solve other use-cases, that's a compelling case to support eggs.

This patch is well-engineered. I'm inclined to accept it.

We probably should update the docs to remove the exclusion of eggs support. @asottile Can you do that?

[jaraco]

In GitLab by @asottile on May 12, 2019, 16:04

added 1 commit

• ef0ce71 - Support .egg installations

Compare with previous version

[jaraco]

In GitLab by @asottile on May 12, 2019, 16:05

done!

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:23

Thanks for that change. Reading that section again, I see that #19 was where that decision was originally made.

Re-reading the conclusion there, I see:

I'm happy to resume the conversation if others arrive here and find that the recommendation is cost-prohibitive or otherwise not suitable, or there's more that importlib_metadata can do (simply) to help accommodate the situation.

I believe this Merge Request resumes that conversation and that the benefits of adding this change far outweigh the costs.

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:23

approved this merge request

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:24

merged

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:24

mentioned in commit d80256c

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:32

mentioned in commit b4af125

[jaraco]

In GitLab by @asottile on May 14, 2019, 13:32

Awesome, thanks! 🎉

[jaraco]

In GitLab by @jaraco on May 14, 2019, 13:35

0.12 should release with this code. I do advise for pytest-pluggy and other projects to try to move away from setup.py test regardless.

[jaraco]

In GitLab by @asottile on May 14, 2019, 13:40

I agree, we're not using it ourselves but our consumers sadly are :(

[jaraco]

In GitLab by @asottile on Jun 18, 2019, 15:40

mentioned in merge request !78