pmxbot IRC Log Viewer

[09:44:47] <pf_moore> Has anything happened with Travis permissions on pip? I'm sure I used to be able to rerun builds, but I don't seem to be able to now :-(

[09:45:19] <pf_moore> PR #1745 looks OK to me but the build's failing - I suspect travis funnies and want to rerun to see if the failures "just go away"

[09:57:45] <dstufft> pf_moore: are you logged in?

[09:58:31] <pf_moore> dstufft: yeah, I wasn't but I clicked "Sign in with github" and still didn't get the "rerun build" button

[09:59:09] <dstufft> pf_moore: that's strange, you should as far as I know

[09:59:11] <dstufft> let me see if I do

[10:00:01] <dstufft> I restarted them

[10:00:31] <pf_moore> Ta. I just went back in and now have a "cancel build" button.

[10:00:39] <pf_moore> Maybe I needed to reload the page

[10:00:52] <pf_moore> Or sacrifice a chicken to it - who knows?

[10:01:40] <dstufft> ~magic~

[10:01:42] <dstufft> or computer

[10:01:44] <dstufft> something like that

[12:59:44] <Ivo> dstufft: if you insert CacheControl, isn't there a whole lot of custom pip cache code to be yanked out as well?

[12:59:55] <dstufft> Ivo: yes

[13:00:01] <dstufft> I didn't bother doing that yet

[13:00:11] <dstufft> we have two different caches in pip

[13:00:16] <Ivo> dstufft: I wanted to see lots of blood red in https://gist.github.com/dstufft/11213903 :/

[13:00:34] <dstufft> we have the in process cache which caches /simple/index/ pages for the life time of the process

[13:00:38] <dstufft> and we have the download cache

[13:02:27] <Ivo> dstufft: could you comment on https://github.com/pypa/pip/pull/1718 ?

[13:05:59] <dstufft> will take a look in a moment

[13:06:50] <Ivo> mainly would love input on how to structure the associated user-facing option

[13:28:17] <dstufft> hmm

[13:28:44] <dstufft> --safe-host localhost --safe-host 127.0.0.1 --safe-host pypi.local

[13:56:06] <dstufft> agronholm: ping

[14:28:59] <simlun> Hi! Is it a good idea to use Warehouse as an in-house Python package index? Devpi has a very nice feature set, but is (as officially stated) lacking a nice web UI. Warehouse's web UI looks very promising.

[14:35:43] <Ivo> simlun: no, it's under heavy development

[14:36:20] <Ivo> unless you're fine with using things considered beta and under heavy development, in which case I'm sure the devs would love beta testers :)

[14:38:06] <DanielHolth> do you need a nice web interface for a mainly-cache-of-pypi?

[14:38:36] <simlun> I'd basically like a nice web UI for devpi

[14:39:27] <simlun> trying to investigate alternatives and get a grip of road maps for the relevant projects in development around the Python community

[14:42:18] <DanielHolth> this is cool but it isn't a pypi https://github.com/perone/stallion

[14:43:31] <simlun> oh, thanks!

[14:46:59] <dstufft> warehouse doesn't actually implement any way to add packages to it yet either

[14:47:05] <dstufft> it just relies on accessing PyPI's database

[14:50:16] <simlun> dstufft: ok!

[17:10:24] <agronholm> dstufft: ?

[17:11:59] <dstufft> agronholm: is with engine.begin() as conn: conn.execute() the way to do things in a transaction the correct way?

[17:12:44] <agronholm> yes

[17:12:52] <agronholm> well

[17:13:02] <agronholm> actually you get a transaction from begin()

[17:13:04] <agronholm> so

[17:13:57] <agronholm> erm, nvm

[17:14:08] <agronholm> yes, you get a Connection

[17:14:22] <agronholm> http://docs.sqlalchemy.org/en/rel_0_9/core/connections.html?highlight=begin#sqlalchemy.engine.Engine.begin

[17:16:29] <dstufft> agronholm: ok cool

[17:17:05] <agronholm> btw I'm still waiting on your 2 cents on the package verification thing

[17:17:22] <dstufft> oops

[17:17:23] <dstufft> lemme go look!

[17:52:24] <yusuket> dstufft: thanks for the feedback on my pr. I left a comment to the same effect, but just want to clarify that the big work items are:

[17:52:31] <yusuket> wrap multiple operation-performing methods in a sql transaction

[17:52:40] <yusuket> create and utilize the "check if arguments match column name in table" method.

[17:52:58] <dstufft> yusuket: ya

[17:53:10] <yusuket> ok cool, will do

[17:53:25] <yusuket> also I still don’t think renaming the package name normalization to “normalize” is very clear

[17:53:45] <yusuket> if you feel strongly about it I’ll switch it, but I worry that that normalize is a bit general of a name

[17:54:15] <yusuket> what else are we planning on normalizing with that method?

[18:02:26] <dstufft> you can call it normalize_project_name if you want

[18:02:35] <dstufft> I just dind't like the normalize_and_Validate

[18:03:38] <yusuket> haha yeah it is a super long name

[18:03:55] <yusuket> sounds good, I don’t mind validation being implicit

[18:09:33] <dstufft> also warehous uses project not package

[18:09:35] <dstufft> for a name

[18:09:43] <dstufft> package is really ambigious

[18:10:08] <dstufft> Project / Release / Distribution

[18:13:02] <DanielHolth> why m2crypto why

[18:13:10] <DanielHolth> much compiling such discomfort

[18:13:39] <Alex_Gaynor> DanielHolth: what crypto stuff do you need from it?

[18:15:11] <DanielHolth> at the moment, nothing thankfully. it's just a transitive dependency that I'm not really using.

[18:15:31] <yusuket> dstufft: I noticed that. Oops, sorry about the naming of that method.

[18:15:49] <DanielHolth> can it do curve22519 dh yet

[18:15:50] <DanielHolth> :-)

[18:16:13] <DanielHolth> Alex_Gaynor btw enjoyed your pycon talks

[18:16:36] <Alex_Gaynor> thanks

[18:17:19] <agronholm> dstufft: so, package verification...?

[18:17:32] <dstufft> agronholm: almost done writing up a response

[18:17:36] <agronholm> ok!

[18:17:45] <DanielHolth> someone should figure out how to do turing-complete pickle and write programs for it

[18:20:12] <Alex_Gaynor> I'm not sure pickle is turing complete. I think it might just be PDA equivilant

[18:20:17] <Alex_Gaynor> (ignoring the "calls random functions" thing

[18:20:38] <DanielHolth> on second thought we could just use python bytecode

[18:20:58] <DanielHolth> or ast truffles to keep it in the food category

[18:31:10] <DanielHolth> It's a gcc thing about __x86_64__ not being defined

[18:41:20] <DanielHolth> heh, rhel's software collections sets XDG_DATA_DIRS

[19:18:35] <mattip> hi. I made progress with issue 488 - pypy, windows, and virtualenv

[19:19:38] <mattip> but as I am new here I thought I would ask for advice

[19:49:34] <mattip> hmm, am I on the wrong channel, I meant to be asking about virtualenv ...

[19:51:18] <agronholm> mattip: if you want to talk about virtualenv's development, you're on the right channel

[19:52:32] <mattip> agronholm: I would like to contribute a fix to issue 448, but not clear to me what I need to do

[19:53:42] <qwcode> dstufft, files is just files, so what's wrong with shutil.move?

[19:54:10] <agronholm> mattip: regarding the actual code changes, or contribution procedures?

[19:54:30] <dstufft> qwcode: nothing wrong per say, but if we introduce a bug where a directory gets into files then shutil.move will move the directory, copy2 will fail instead

[19:54:32] <mattip> actual code changes, I understand pull requests and pep8 and such

[19:54:56] <dstufft> I think it's better to fail explicitly in that case, than to possibly reintroduce this bug subtly

[19:55:02] <agronholm> well if you don't know what to do about the code then you can't contribute?

[19:55:47] <mattip> here is my question - pypy needs another directory copied, somewhere in install_python()

[19:56:26] <mattip> on windows, with no symlink, the missing directory crashes the first run of the executable

[19:57:01] <mattip> but it seems that install_python() goes to quite some effort to only install a bare minimum std

[19:57:27] <qwcode> dstufft, I guess so, but it's just an os.walk routine that's taking the files list as is and working with them. adding that precaution seems like too much caution.

[19:57:28] <mattip> so I do not want to break that model by copying too much

[19:57:44] <dstufft> qwcode: is there any reason _not_ to use copy2?

[19:58:02] <mattip> s/std/stdlib/

[19:58:31] <qwcode> dstufft, no, it's just not really related to the change. but I'll add it. : )

[19:58:48] <dstufft> :D

[19:59:13] <mattip> can I just submit a patch that copies the whole subtree in install_python() ?

[20:00:48] <dstufft> mattip: you're dealing with some of the heavy magic of virtualenv, I'd have to look over the code to know for sure and I don't have time ATM

[20:02:39] <mattip> np, I'll commit something and wait for comments.

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Wednesday the 23rd of April, 2014