Log file Viewer

[02:31:33] <ErikRose> dstufft: jezdez: ^ There's my hashing stuff, for your reviewing pleasure!

[12:42:47] <Ivo> dstufft: would like to get a release of virtualenv out and then work on fixing it for win py35

[12:43:11] <dstufft> Ivo: I can cut a release in a little bit

[12:46:59] <Ivo> https://github.com/pypa/virtualenv/compare/a2292600206e7307b06fde6b8f59a82ac3477910...develop#files_bucket is compare of my changes for it, if you are interested

[13:00:44] <Ivo> dstufft: you might be interested, i wrote a tool, pipa, as a Minimum Viable Product for a simple HTTPS-enabled self-hosted pypi server https://github.com/Ivoz/pipa

[13:01:12] <Ivo> seeing the page you were writing on using one of those

[13:02:31] <Ivo> Only limitation is it's py3 only, i guess

[13:21:02] <ronny> Ivo: how does it compare to devpi?

[13:23:40] <Ivo> ronny: designed to be simple to setup, and work with self-signed ssl out-of-the-box

[13:25:55] <Ivo> afaik devpi relies on setting up another webserver with https proxy in order to do https

[13:28:31] <Ivo> it also supports "native" setup.py upload

[13:31:55] <ronny> Ivo: devpi also supports setup.py uploads

[13:32:20] <ronny> Ivo: also devpi suppors standalone running, but its kidna sane to let a real webserver handle files

[16:04:18] <ErikRose_> dstufft: Thanks for the initial review-ette. :-)

[16:04:39] <ErikRose_> Does anybody run the pip tests in parallel?

[16:04:56] <dstufft> travis does

[16:05:14] <dstufft> it's a good way to make your computer totally unusable too

[16:05:21] <ErikRose_> Sounds great.

[16:05:22] <dstufft> if you ever wanted to do that for 10 minutes or so

[16:05:35] <ErikRose_> Beats waiting 40 minutes for py27 and py33!

[16:06:28] <dstufft> I rarely bother running the tests locally tbh

[16:06:34] <dstufft> unless one in particular is failing

[16:06:38] <dstufft> then I'll run _just_ that one

[16:07:03] <ErikRose_> I haven't yet figured out how to run a single test on a specific version of Python that my main venv isn't.

[16:07:35] <ErikRose_> Ah, tox --.

[16:07:44] <ErikRose_> Elegant.

[16:08:48] <dstufft> yea

[16:09:03] <dstufft> you can use tox -e py33 -- -k <name of test>

[16:09:14] <dstufft> or a path to a particular file instead of -k <name of test>

[16:09:15] <dstufft> or w/e

[16:09:15] <ErikRose_> Yep, I did discover that one.

[16:09:49] <ErikRose_> Hey, this is nice.

[16:15:36] <dstufft> ErikRose: fwiw I sort of wonderif -H should be a shortened form of --require-hashes... but then I'm not sure how many people are going to be typing --require-hashes on the CLI either, seems like something they'd put in their deployment scripts or something but otherwise not something people would interact with

[16:16:00] <ErikRose> Yes, I don't think --require-hashes will be used much explicitly.

[16:16:19] <ErikRose> It's the sort of thing a sysadmin might throw into a script to enforce that the committers do the right thing.

[16:17:43] <dstufft> I'm kind of loathe to use short options unless I think it's going to be something people are typing a lot, just because we only have 52 of them, and the more of them we have the more confusing things become when looking at a particular command invocation

[16:18:07] <ErikRose> I don't feel strongly about it. We can start with --hash and see if it becomes annoying.

[16:23:25] <ErikRose> I should use more backslash continuations in my requirements tests. I think that's what people will mostly do.

[18:54:19] <Ivo> ronny: standalone running with HTTPS?

[19:02:36] <ronny> Ivo: im not 100% sur on the https bt it should be doable in a few lines of code

[19:30:03] <ErikRose> Btw, I suspect that the tox test failures are triggered by running in parallel. I don't see them locally--at least not always.

[19:35:20] <ErikRose> dstufft: You should upgrade your computer; -n 8 on my 2012 rMBP gets the tests done in 5m, machine usable all the while!

[19:35:55] <ErikRose> Another interesting tidbit: all the tests passed, even at -n 8, even the ones that fail on Travis. Whee.

[19:42:13] <dstufft> ErikRose: I haven't tried it on my new iMac tbh

[19:42:20] <dstufft> my new one might be better idk!

[19:42:22] <ErikRose> You may be pleasantly surprised. :-)

[19:42:31] <dstufft> I hope the 5k is better than my 2011 era

[19:42:43] <ErikRose> It seems likely!

[21:08:56] <Ivo> ronny: ahh, but does that few lines of code generate a new self-signed SSL cert for you straight away, or are you going to have to go reading ssl x509 manpage docs for 10 minutes again? :)

[22:40:15] <ErikRose> Heh, I ran py.test -n 8 on my laptop, and it did an emergency power-off in 10 seconds. It had reported 1:30 left in the battery just before.

[22:48:18] <ErikRose> py.test seems to be of the loud/louder/loudest school of visual design.

[22:49:58] <dstufft> ErikRose: what do you mean?

[22:50:30] <ErikRose> I'm just laughing at all the racket. Why say "_________" when you could say "==========="? Or, better still, "!!!!!!!!!!!!!!!!!!!!!!!"! :-)

[22:50:48] <ErikRose> I need to port over nose-progressive.

[23:18:49] <ErikRose> I wonder why a bunch of my "pip search" tests are suddenly failing, even on commits that passed yesterday.

[23:19:33] <ErikRose> Wait, these are all network tests.

[23:19:47] <ErikRose> dstufft: Are you doing something to the search services?

[23:20:01] <dstufft> on PyPI? No

[23:20:11] <ErikRose> hmm

[23:20:49] <ErikRose> Anyway, I think I've nearly made this --no-deps implication go away. I just have to make sure the tests aren't failing any worse than before.

[23:27:49] <ErikRose> dstufft: Alright, I've removed the --no-deps implication. It was surprisingly not bad.