[issue27425] Tests fail because of git's newline preferences on Windows

Ned Deily added the comment: OK, based on Steve's response, I am closing this again as it sounds like it should not be an issue going forward with releases and also not for dev builds as long as people do a fresh clone if necessary. Further discussion on hardening the tests can take place

[issue27425] Tests fail because of git's newline preferences on Windows

Terry J. Reedy added the comment: There is still the problem with test_winreg freezing CommandPrompt, #30715. For me, this is the worst test bug ever. -- ___ Python tracker

[issue27425] Tests fail because of git's newline preferences on Windows

Steve Dower added the comment: There's nothing that needs to block the release as far as I'm concerned. The main fix was that I needed a fresh clone on my build machine to pick up the gitattribute changes. The other bug is still open to enhance some tests and skip one in the release tree,

[issue27425] Tests fail because of git's newline preferences on Windows

Ned Deily added the comment: So do we have a resolution or resolutions for this yet? And is bpo-30716 truly a duplicate? If so, let's use one or the other. -- ___ Python tracker

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: I reproduce the bug after installing Python 3.6.2rc1 on Windows. I reopen the issue and set the priority to release blocker. -- nosy: +benjamin.peterson, larry, ned.deily priority: normal -> release blocker resolution: fixed -> status: closed -> open

[issue27425] Tests fail because of git's newline preferences on Windows

Terry J. Reedy added the comment: Installed 3.6.2rc1 is showing the same errors for test_random and test_sax. #30716. -- nosy: +terry.reedy ___ Python tracker

[issue27425] Tests fail because of git's newline preferences on Windows

Terry J. Reedy added the comment: Why de we would 'want' /r/n on Windows? What beside Notepad can't handle /n? -- ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: I'm happy to see this "old" issue now fixed :-) -- ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: It still might be nice to adjust some of the tests to not care about how line endings are checked in, but I think we're in a good enough place now with .gitattributes that we can close this issue. Any other changes can be done in a new issue. --

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: New changeset 5fe8ac69f99d58232a5234cfca78f05f7223a782 by Zachary Ware in branch '2.7': [2.7] bpo-27425: Be more explicit in .gitattributes (GH-840) (GH-2086) https://github.com/python/cpython/commit/5fe8ac69f99d58232a5234cfca78f05f7223a782 --

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: > Zach, do you want to update Python 2.7 for your latest change as well? PR 2086 is in for that, but it's a huge diff due to line endings changing. As noted on the PR, I want Benjamin to give his OK before I merge it. --

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: Zach, do you want to update Python 2.7 for your latest change as well? -- ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: New changeset afa6a389606402779c5048a7211edbf2678cecff by Zachary Ware in branch '3.5': [3.5] bpo-27425: Be more explicit in .gitattributes (GH-840) (GH-2084) https://github.com/python/cpython/commit/afa6a389606402779c5048a7211edbf2678cecff --

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: New changeset 964c261dc9a6a901f50d5596d88248bfc4251a55 by Zachary Ware in branch '3.6': [3.6] bpo-27425: Be more explicit in .gitattributes (GH-840) (GH-2083) https://github.com/python/cpython/commit/964c261dc9a6a901f50d5596d88248bfc4251a55 --

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Zachary Ware : -- pull_requests: +2150 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Zachary Ware : -- pull_requests: +2148 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Zachary Ware : -- pull_requests: +2147 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

Zachary Ware added the comment: New changeset 6b6e68776663c0dda04b6a36609297728da2ae9e by Zachary Ware in branch 'master': bpo-27425: Be more explicit in .gitattributes (GH-840) https://github.com/python/cpython/commit/6b6e68776663c0dda04b6a36609297728da2ae9e --

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: Wait, why is this issue still open? Tests now pass on Windows on all branches. The issue can be closed, except if I missed something? -- ___ Python tracker

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Mariatta Wijaya : -- versions: +Python 3.7 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: New changeset cf57fe13b44e3994096ae79f32c11475f333a94f by Victor Stinner in branch '3.5': bpo-27425: Add .gitattributes, fix Windows tests (#844) https://github.com/python/cpython/commit/cf57fe13b44e3994096ae79f32c11475f333a94f --

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by STINNER Victor : -- pull_requests: +741 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Zachary Ware : -- pull_requests: +737 ___ Python tracker ___ ___

[issue27425] Tests fail because of git's newline preferences on Windows

Steve Dower added the comment: More likely it needs a fresh clone to fix up those files, unless they've changed since the attributes file was added. -- ___ Python tracker

[issue27425] Tests fail because of git's newline preferences on Windows

STINNER Victor added the comment: In the master branch, .gitattributes now contains: --- *.pck binary Lib/test/cjkencodings/* binary Lib/test/decimaltestdata/*.decTest binary Lib/test/sndhdrdata/sndhdr.* binary Lib/test/test_email/data/msg_26.txt binary Lib/test/xmltestdata/* binary

[issue27425] Tests fail because of git's newline preferences on Windows

Emanuel Barry added the comment: Martin: Indeed, seems like it's backwards for some reason. I'm not sure what happened when I regenerated the index; I removed the patches now anyway. I think the .gitattributes patch would be fine to go on its own. It's my understanding that line endings don't

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Emanuel Barry : Removed file: https://bugs.python.org/file44220/fix_newlines_1.patch ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Emanuel Barry : Removed file: https://bugs.python.org/file44225/fix_newlines_2.patch ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Martin Panter added the comment: Emanuel: fix_newlines_2.patch seems to change the contents of batch from CRLF to LF newlines, undoing revision 640ccb924b5f. This seems like a step in the wrong direction. Notice $ means LF, and ^M$ means CRLF: $ curl

[issue27425] Tests fail because of git's newline preferences on Windows

Jim Jewett added the comment: I think if the test is ensuring one line-ending type, then there should be another test ensuring the other... but I think it would be best if there were a single test file that had both types of line-endings, just to ensure that the code doesn't cheat by reading

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Emanuel Barry : Added file: https://bugs.python.org/file44225/fix_newlines_2.patch ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Emanuel Barry added the comment: Whoops, turns out I was using the wrong approach for binary files. Here come add_gitattributes_2.patch and fix_newlines_2.patch (thanks Zachary for pointing this out). -- Added file: https://bugs.python.org/file44224/add_gitattributes_2.patch

[issue27425] Tests fail because of git's newline preferences on Windows

Emanuel Barry added the comment: Here are three patches for this: - add_gitattributes_1.patch adds `.gitattributes`, with all the newline preferences in `.hgeol` (with the exception of the "native" rule at the end, which in git is the default newline preference). - fix_newlines_1.patch fixes

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Emanuel Barry : Added file: https://bugs.python.org/file44221/test_random_warning_1.patch ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

Changes by Emanuel Barry : Added file: https://bugs.python.org/file44220/fix_newlines_1.patch ___ Python tracker ___

[issue27425] Tests fail because of git's newline preferences on Windows

R. David Murray added the comment: Yes .gitattributes looks like the correct solution. The problem is that most of the text files we *want* git to transform per platform, it's just a few that we don't. Mercurial actually added support for this for us when we switched to it. --

[issue27425] Tests fail because of git's newline preferences on Windows

Emanuel Barry added the comment: I agree, but I wouldn't expect my line ending preferences to make tests fail. I think that with the move to GitHub already underway, it's not unreasonable to (at least consider to) add a .gitattributes file. Added Brett to nosy since he's responsible for PEP

[issue27425] Tests fail because of git's newline preferences on Windows

Martin Panter added the comment: If it were me, I might try to use a version of Git that doesn’t mess with the files. But if this becomes a common problem (esp with the move from Mercurial to Git), maybe we can add a “.gitattributes” file . We aleady