Log file Viewer

[13:16:35] <herrwolfe45> msabramo: were you asking about my PR because you wanted to close it, or because you wanted to work on it (this was the PR related to a warning for extras-require)?

[13:18:02] <herrwolfe45> msabramo: this is PR https://github.com/pypa/pip/pull/2142

[13:26:38] <xafer> https://github.com/pypa/pip/pull/2153 would love to be in pip 6.0 :o

[15:51:25] <msabramo> herrwolfe45: Neither. It seems like it could potentially be a good change and I thought maybe you had forgotten about it. I guess I would say it should be closed if you weren't planning to work on it but since you are, that seems cool

[15:52:35] <msabramo> and it would be nice to close issues (whether by completing them or closing things that people aren't working on) since there are 378 open issues, which is a bit unwieldy

[15:58:26] <herrwolfe> msabramo: sorry about my name change - didn't see your message

[15:58:45] <msabramo> herrwolfe: no worries

[15:59:10] <herrwolfe> msabramo: I see what you mean about having open issues. I am hoping I can get some sort of a fix for this completed by friday

[15:59:21] <herrwolfe> msabramo: after that I will be on vacation for a week

[15:59:39] <msabramo> herrwolfe: cool

[15:59:39] <herrwolfe> msabramo: but I'd definitely like to complete it

[19:45:01] <dstufft> qwcode: two things, can we switch www.pypa.io to use HTML Dir instead of Html? and https://www.pypa.io/en/latest/peps.html the PEP 440 thing probably should mention packaging too

[19:49:21] <tomprince> https://www.pypa.io/en/latest/peps.html#pep453-explicit-bootstrapping-of-pip-in-python-installations could mention 2.7.9.

[19:50:02] <tomprince> And should https://www.pypa.io/en/latest/peps.html#pep438-transitioning-to-release-file-hosting-on-pypi mention 470?

[20:01:29] <qwcode> dstufft, yes, will do on both counts

[20:18:18] <xafer> any feedback on https://github.com/pypa/pip/pull/2153 ?

[21:08:31] <qwcode> tomprince, the PEP470 section mentions that it reverts PEP438. I'll just remove the PEP438 section

[21:09:35] <tomprince> qwcode: Well, 470 hasn't been accepted yet. I had thought perhaps something like the note in 425/427.

[21:10:10] <qwcode> tomprince, yea, that sounds good

[21:44:46] <dstufft> jaraco: does my explanation on >=,<= and > and < make sense?

[21:45:06] <jaraco> dstufft, I'm not getting it on first read.

[21:45:27] <dstufft> <V and >V imply a !=V.*

[21:45:30] <jaraco> Your explanation made sense except that it seemed to contradict the explanation before.

[21:46:02] <dstufft> what part contradicts do you think? I'm not always the greatest at putting what makes sense in my head into words

[21:47:17] <jaraco> In the OP, '1.0.1' in Requirement.parse('foo>1.0')

[21:47:38] <dstufft> >1.0 implies !=1.0.*

[21:47:52] <jaraco> Oh!

[21:48:03] <dstufft> exclusive operations are a little wierd for >

[21:48:12] <jaraco> Yes, now I get it.

[21:48:12] <sigmavirus24> huh

[21:48:17] <dstufft> but I didn't see very many uses of > in general

[21:48:20] <dstufft> mostly >= and <

[21:48:54] <jaraco> !schneier dstufft

[21:48:54] <pmxbot> dstufft knows Chuck Norris' private key.

[21:49:04] <dstufft> lol

[21:49:35] <dstufft> this is one of those rules that probably fixed a number of bugs and problably created a few bugs too

[21:49:51] <jaraco> It's going to take a bit of getting used to.

[21:50:05] <jaraco> But more and more, it feels like the right move.

[21:50:48] <dstufft> I'm glad that it does :D I certainly think it is, I've probably spent more time with PEP 440 than anyone so I've gotten used to thinking that way already

[21:51:04] <dstufft> it's really the idea that version specifiers aren't really just less than and equal operators

[21:52:04] <sigmavirus24> dstufft: so >1.0.0 would allow for 1.0.1 then?

[21:52:11] <dstufft> sigmavirus24: yes

[21:52:22] <sigmavirus24> but not 1.0.0.1 =P

[21:52:25] <dstufft> correct

[21:52:27] <sigmavirus24> that makes sense now

[21:52:47] <sigmavirus24> I was slightly surprised at first when you said it didn't allow for 1.0.* but I get it

[21:53:00] <dstufft> it makes more sense when you consider the < use case

[21:53:19] <dstufft> <3 excludes !=3.*, which means <3 doesn't match 3.dev0

[21:53:25] <dstufft> in PEP 440

[21:53:39] <dstufft> but it _does_ match 3.dev0 in setuptools pre 8

[21:53:45] <sigmavirus24> I thought <3 meant you loved me =P

[21:53:58] <sigmavirus24> dstufft: yeah that makes a lot more sense

[21:54:00] <dstufft> which is why you see things like <3dev or <=2.99

[21:54:05] <sigmavirus24> yeah

[21:55:26] <dstufft> now in a post PEP 440 world, you'd really write something like >=2.0,<3 as ~=2.0

[21:55:44] <dstufft> or >=2.1,<3 as ~=2.1

[21:56:18] <dstufft> but that's not valid in setuptools before 8.0, so it'll require time till people can rely on that syntax

[21:58:37] <sigmavirus24> yeah

[21:58:47] <sigmavirus24> I like that from gems (~> 2.1)

[22:05:19] <dstufft> the one downside with ~= is that you can't really depend on a micro version

[22:05:33] <dstufft> if you do ~=2.1.3 then you're limited to 2.1.x

[22:07:04] <dstufft> npm has ^ to handle that

[22:07:16] <dstufft> ^2.1.3 is >=2.1.3,<3

[22:07:56] <dstufft> I have some ideas for a next gen PEP 440 (which just adds things, doesn't change anything that's here now), but this is a good step forward I think

[22:15:55] <sigmavirus24> yeah it's a really good step forward

[22:16:10] <sigmavirus24> !m dstufft

[22:16:10] <pmxbot> you're doing good work, dstufft!

[22:16:48] <dstufft> next gen PEP 440 will probably include "and" and "or" and parenthesis for grouping

[22:17:18] <dstufft> (in addition to ,)

[22:17:38] <dstufft> modeling that is going to be "fun" though

[22:18:22] <Arfrever> dstufft: Also for different packages?

[22:18:42] <dstufft> jaraco: currently PEP 440 states that release candidates get normalized as 1.0c1, I want to change this to 1.0rc1... There's not really a process for ammending a PEP like that, but Nick said if pip and setuptools buy in to it "whose gonna say we can't jsut amend it"

[22:18:54] <dstufft> jaraco: so are you OK with switching it so 1.0rc1 is the preferred form?

[22:19:18] <dstufft> Arfrever: that's a different (but similar) problem, I'd like to expand more on them too!

[22:19:32] <Arfrever> dstufft: E.g. dependency on "PyQt4 or PySide"

[22:20:10] <dstufft> jaraco: (to be clear, both 1.0c1 and 1.0rc1 are currently accepted, and will continue to be, this is just about changing the "normalized" or preferred form)

[22:20:25] <dstufft> Arfrever: yea

[22:20:30] <dstufft> Arfrever: it's something i'd like to do

[23:03:28] <jaraco> dstufft: sounds fine to me.