Repository navigation
Only run CI checks when appropriate files have changed #14
Description
Activity
- added a commit that references this issue
on Feb 14, 2017 This is now done. @brettcannon could you please close this?
Side note: Should we allow "Core developers" team to close issues on this repo?
Closed and now the core team has write permissions (just simply forgot to add the team).
And thanks for the change! I'm sure not only PR submitters appreciate the faster turn-around but I bet Travis appreciates us using less free resources. 😄
Reacted by Berker PeksagI don't think the Doc only test works. See this PR here: python/cpython#3072
The travis run: https://travis-ci.org/python/cpython/builds/263366915 (and it's console log)It triggered a full code rebuild and test even though the only thing that has changed is
Doc/library/socket.rstIt was working but it seems something has changed such that it didn't succeed in this case.
I've done some investigating and this is due to how Travis gets the merged code from Github.
Following the sequence of commands it used to get the repo here, it went:
git clone --depth=50 https://github.com/ammaraskar/cpython.gitgit fetch origin +refs/pull/5/mergegit checkout -qf FETCH_HEAD
I replicated these commands locally and ran
git logand the problem became obvious. Github in their specialrefs/pull/*/mergebranch creates a merge commit which means at that point this git log looks something like:commit 472056eb60b0601dcded798229bebbddfd307945 (HEAD) Merge: b072c6d0c6 86cece89d5 Date: Fri Aug 18 05:43:43 2017 +0000 Merge 86cece89d5c6452a92004af0276147187034d41f into b072c6d0c69849690ff03c0ce4f1911139b02ce6 commit b9ff498793611d1c6a9b99df464812931a1e2d69 (master) Author: Elmar Ritsch <35851+elritsch@users.noreply.github.com> Date: Fri Aug 18 05:23:51 2017 +0200 Fix broken `Show Source` links on documentation pages (GH-3113) The `Show Source` was broken because of a change made in sphinx 1.5.1 In Sphinx 1.4.9, the sourcename was "index.txt". In Sphinx 1.5.1+, it is now "index.rst.txt" ... EVERY SINGLE COMMIT TO MASTER ON A TIME AFTER MY PR COMMIT DATE ... commit 86cece89d5c6452a92004af0276147187034d41f Author: Ammar Askar <ammar@ammaraskar.com> Date: Sat Aug 12 15:15:07 2017 -0400 Fix Linux availability version for SO_PEERSECAnd so the
git diffpicks up on all the commits in the middle because the PR commit ends up in a really old slot in history.OK, so it's basically it's imperfect but isn't hurting anything. Thanks for the detailed report, @ammaraskar ! I'm going to close this again and we can just live with occasionally building more than necessary.
There's a way around it, I opened up a PR to fix it: python/cpython#3129
Whether it's worth the added complexity to the build process, that I don't know.
- added a commit that references this issue
on Jan 26, 2018 - added a commit that references this issue
on Jan 27, 2018
E.g. don't run code coverage or the test suite if only the docs changed (or vice-versa).