Log file Viewer

[18:34:42] <nedbat> what makes this message, and what would I have to change to make it stop: Could not find a tag or branch 'b1f36c0c6f515b7c0d39744a46661d25cfed89cd', assuming commit.

[19:21:18] <pjdelport> nedbat: https://github.com/edx/XBlock/blob/b1f36c0c6f515b7c0d39744a46661d25cfed89cd/setup.py ?

[19:21:27] <pjdelport> That's probably versioneer getting confused by something.

[19:23:20] <nedbat> pjdelport: how did you find the commit?

[19:23:30] <pjdelport> Google :)

[19:23:46] <nedbat> pjdelport: i believe this message happens even without versioneer involved, but I could be wrong.

[19:24:06] <pjdelport> nedbat: Versioneer is probably invoking git to try and get the version for the commit.

[19:24:29] <dstufft> I think that's from pip, or maybe git itself

[19:24:37] <nedbat> pjdelport: i continue to believe that versioneer is not always involved. :)

[19:25:42] <dstufft> it must come from git itself

[19:26:09] <dstufft> searching pip finds a test that asserts we get that output, but I don't see any code writing that output

[19:26:17] <dstufft> https://github.com/pypa/pip/search?utf8=%E2%9C%93&q=Could+not+find+a+tag+or+branch

[19:27:31] <nedbat> dstufft: https://github.com/pypa/pip/blob/91172f578a6bc548470e8d6a750e2c4b3ae6ff4a/pip/vcs/git.py#L79

[19:28:40] <dstufft> hm

[19:28:44] <dstufft> wonder why it didn't come up in search

[19:28:46] <dstufft> oh well

[19:28:50] <dstufft> guess it's our fault

[19:33:49] <dstufft> I had to try real hard to find out where that got introduced

[19:33:51] <dstufft> it's quite old

[19:33:52] <dstufft> https://github.com/pypa/pip/commit/a8de40cac1efc1e64dff6e62ad6939499027ed0d

[19:36:16] <dstufft> nedbat: probably if you want I think that can probably be just removed from pip's codebase, or at the very least switched to a debug

[19:37:10] <nedbat> dstufft: i searched all of github for that string, and it said there were no matches at all...

[19:37:51] <dstufft> nedbat: computer.

[19:38:01] <nedbat> dstufft: I'd remove it in the case that rev matches [0-9a-fA-F]{40} or whatever the right SHA match is.

[19:38:29] <dstufft> eh, it's kinda a crummy warning even for shorter hashes

[19:39:54] <nedbat> dstufft: i thought the point was that tags and hashes are ambiguous for short hashes

[19:40:38] <nedbat> dstufft: rev = "bed123"

[19:41:07] <dstufft> nedbat: git will error if it's ambiguous

[19:41:17] <dstufft> I'm not sure why we need to warn for something git doesn't

[19:41:19] <nedbat> oh really?

[19:41:31] <dstufft> yea, you can use the shortest possible string that is unique

[19:43:30] <dstufft> $ git checkout 32b

[19:43:30] <dstufft> error: pathspec '32b' did not match any file(s) known to git.

[19:43:43] <dstufft> $ git checkout 32b4

[19:43:43] <dstufft> Note: checking out '32b4'.

[19:43:59] <dstufft> the message for git checkout isn't great, for things that don't also support paths it's better:

[19:44:20] <dstufft> $ git show 32b

[19:44:20] <dstufft> fatal: ambiguous argument '32b': unknown revision or path not in the working tree.

[19:47:52] <xafer> switching to a debug seems sensible

[19:50:11] <dstufft> we could also possible use rev-parse to figure it out instead of using show-refs to get all of the branches and tags

[19:50:17] <dstufft> not sure why it's doing that to begin with

[19:50:47] <dstufft> I guess to see if it needs to add origin/ to the front?

[20:50:41] <_habnabit> is it possible/sensible to pin to an upstream version in install_requires? something like `~=0.1.0,>=0.1.0+upstream.2`

[20:51:04] <_habnabit> wasn't sure what the semantics of `~=0.1.0+upstream.2` would be

[20:52:48] <_habnabit> hm, looks like `~=0.1.0,>=0.1.0+upstream.2` doesn't actually look at the bit after the +

[20:53:54] <_habnabit> even just `~=0.1.0,!=0.1` is filtering out `0.1+upstream.2`

[21:31:44] <dstufft> _habnabit: Um, I think !=0.1 will filter it

[21:33:55] <tdsmith> is pip search throwing xmlrpc errors for everyone?

[21:34:28] <mathieui> yes.

[21:34:55] <dstufft> computers are the worst

[21:35:16] <tdsmith> concur

[22:07:14] <_habnabit> dstufft, !=0.1 filtered out 0.1+upstream.2

[22:17:18] <agronholm> dstufft: how's the fixing going with pypi?