pmxbot IRC Log Viewer

[00:19:10] <dstufft> lifeless: um, I'm not sure offhand. I think it's less there will be actual build failures and more wheels will just silently do the wrong thing

[00:19:28] <lifeless> dstufft: do you want me to add a blacklist feature?

[00:19:43] <lifeless> httppretty 0.8.0 fails to build, so I can use that as a test

[00:20:09] <dstufft> lifeless: as in, "don't try to build wheels for X projects"? that would probably be useful for people to be able to flag some things as not working with wheel caching but still get the basic overall feature

[00:21:32] <dstufft> lifeless: sorry, been playing bloodborne heh

[00:21:55] <dstufft> -> goes back to playing that

[00:22:37] <lifeless> dstufft: :)

[00:42:57] <lifeless> ahh

[00:43:01] <lifeless> wheelbroken-0.1

[20:00:14] <lifeless> morning :)

[20:04:35] <DanielHolth> hello

[21:05:43] <lifeless> dstufft: how does one re-try a build? the py26 failure on 2616 is a network timeout

[21:05:58] <dstufft> lifeless: with travis you have to have commit rights to do that

[21:06:01] <dstufft> it's not the greatest

[21:06:03] <dstufft> sec

[21:06:34] <dstufft> lifeless: restarted it

[21:07:21] <lifeless> thanks

[21:14:21] <lifeless> dstufft: there are some junk branches in github's pip repo

[21:14:35] <lifeless> dstufft: like issue-381-do-not-use-pwd-in-installation-process

[21:14:43] <dstufft> yea

[21:14:54] <dstufft> I've been meaning to look at them and figure out if we can get rid of them

[21:15:01] <dstufft> just haven't bothered

[21:15:11] <lifeless> dstufft: tey're messing up my tab complete :)

[21:15:23] <lifeless> dstufft: what do you need to know vis-a-vis 'can' ?

[21:16:01] <dstufft> lifeless: just need to make sure there isn't anything in them that we'd want to save. My guess is "no" but

[21:16:05] <dstufft> doesn't hurt to check

[21:16:09] <lifeless> dstufft: ok

[21:16:24] <lifeless> so issue-381 - do this - git log develop..38d7f5a69f3d5d07dc32a9e939b3b04d0262e9ed

[21:16:32] <lifeless> there are 4 commits

[21:16:35] <lifeless> from 2011

[21:17:33] <lifeless> dstufft: (thats the only one glitching me, because I have issue- named branches too :)

[21:18:05] <dstufft> hm

[21:18:07] <dstufft> actually

[21:18:11] <dstufft> maybe I did delete them

[21:18:36] <lifeless> dstufft: if you deleted it locally, you need to do git push $repo --delete $name to make it stick

[21:18:38] <dstufft> I don't see them here https://github.com/pypa/pip/branches/all

[21:19:08] <lifeless> oh

[21:19:11] <lifeless> git I had you

[21:19:17] <lifeless> hate

[21:19:38] <lifeless> there's still a revert-2547-revert-2513-ssh_and_commit_support

[21:20:48] <dstufft> I can't delet eit because an open PR depends on it

[21:20:53] <lifeless> ah

[21:21:12] <dstufft> looks like someone made a PR against the wrong branch

[21:21:16] <dstufft> commented

[21:21:47] <lifeless> ok 2616 passed CI

[21:59:53] <lifeless> dstufft: what are your thoughts on the setup.cfg's in my topo pr?

[22:00:39] <dstufft> lifeless: you mean whether to include them or not?

[22:00:42] <lifeless> yeah

[22:01:09] <lifeless> https://github.com/pypa/pip/pull/2616#discussion-diff-27392981

[22:01:34] <dstufft> I don't have a strong preference either way, I have a slight preference for what qwcode said and given he said it too it probably is best to remove it for now and re-add it in the other PR

[22:01:44] <lifeless> k

[22:01:55] <lifeless> that, the docstring, and by implication the docs. Anything else?

[22:02:13] <dstufft> I'm looking over it now, was eating some pizza

[22:02:19] <lifeless> :)

[22:02:25] <lifeless> what time is it now for you?

[22:09:41] <dstufft> 6pm

[22:09:48] <lifeless> wow 7h out now

[22:10:33] <dstufft> Hm

[22:10:42] <dstufft> are you in the same timezone as brisbane

[22:10:58] <dstufft> I don't know the relative locations on that side of the world very well

[22:11:06] <lifeless> NZ

[22:11:29] <lifeless> except for a couple of weeks where we do DST differently, 2 hours east of sydney

[22:11:32] <dstufft> what's a city in NZ

[22:11:44] <lifeless> christchurch

[22:11:54] <lifeless> auckland is usually what you'll find on tz charts

[22:12:09] <dstufft> added to my world clock sidebar thing

[22:13:09] <dstufft> I try to keep clocks in the areas for people I end up talking to a lot to make it easier to time things heh

[22:13:27] <lifeless> yeah

[22:23:31] <dstufft> lifeless: I'm trying to gauge the side effect of this parent_req thing, is this going to make it so that what used to be a double requirement error is no longer an error?

[22:28:26] <lifeless> dstufft: no

[22:28:37] <lifeless> dstufft: there were two calls of add_requirement

[22:28:41] <lifeless> one is the UI layer

[22:28:46] <lifeless> which passes parent_req=None

[22:28:55] <lifeless> note that that is in the updated guard for the double requirement

[22:29:10] <lifeless> the second was req_set itself

[22:29:27] <lifeless> and note that all prior calls within req_set were guarded by 'if not self.has_requirement'

[22:29:42] <lifeless> even unnamed->named resolution

[22:29:45] <dstufft> ah

[22:29:52] <dstufft> that last bit was the part I was missing

[22:31:30] <dstufft> lifeless: Ok, then I think I feel good about this PR

[22:31:51] <dstufft> I also see some potenttional improvements it opens up to the handling of multiple things depending on the same thing

[22:32:09] <dstufft> but I don't want to do too much in this PR to add those

[22:32:19] <lifeless> yeah. I'm skeptical about the improvements there

[22:32:24] <lifeless> most folk don't have multiple spindles

[22:32:57] <lifeless> making each build internally parallelise to the machines capacity would be better than doing more mostly-IO-bound work from within pip

[22:33:01] <lifeless> but thats a different discussion

[22:33:58] <dstufft> lifeless: I mean, two things depending on "six"

[22:34:11] <dstufft> right now you know we only take the first one added to the requirement set into consideration

[22:34:34] <dstufft> a future PR to this could change the one branch so that instead of just pulling the original requirement out of the dictionary, we combine them

[22:34:52] <lifeless> oh, you mean the requirements specification

[22:34:59] <lifeless> so we'd and together the relations

[22:35:05] <lifeless> yes, this would be where that happens

[22:35:23] <lifeless> I agree. There should be an issue on that already ?

[22:35:40] <dstufft> I think there is yea, it might be under "we need a proper dep solver"

[22:35:42] <lifeless> I'm just working on updating ReqSet with the built wheels atm and then 2140 will be ready

[22:35:51] <lifeless> I don't like that bug, its too broad

[22:36:07] <lifeless> like, success isn't defined, and we wouldn't want a single PR claiming to do it all at once

[22:36:45] <dstufft> yea, a lot of our bugs are that way sadly :[

[22:37:36] <lifeless> solutions shouldn't be issues, IMNSHO

[22:37:41] <lifeless> symptoms

[22:37:45] <lifeless> desires

[22:37:47] <lifeless> features

[22:38:02] <lifeless> 'please stop shitting all over me in form X' :)

[23:27:05] <dstufft> lifeless: did you mean to commit an egg-info in 2616?

[23:32:39] <lifeless> dstufft: no, fixing

[23:32:41] <qwcode> lifeless, "that bug, its too broad". for now at least its gives us something to dupe against, and collect ideas

[23:33:41] <qwcode> we get a lot of "my dependencies didn't resolve as expected" tickets

[23:34:07] <lifeless> dstufft: fixed

[23:35:55] <lifeless> qwcode: I can see the value in collecting ideas; duping different symptoms into one issue when they may-or-may-not be solved separately tends to have bad side effects IME

[23:39:24] <qwcode> lifeless, if you have a clear idea how to break that apart, go for it. but pip's "first found, wins" logic seems like a discrete tangible thing that we want to replace

[23:41:08] <lifeless> qwcode: I've enough things juggling atm; my whinge was more abstract than I think we need to rework the bug tracker data for it right now

[23:41:25] <lifeless> qwcode: but thanks, if I get the activation energy up I may do something

[23:41:34] <lifeless> first, lunch

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Tuesday the 31st of March, 2015