| 07:51 | <annevk> | Every other week there is this tweet: https://twitter.com/sergiolujanmora/status/554072564989837313 I wonder where they originate... |
| 11:05 | <gsnedders> | jgraham: do you have any opinions on py.test v. nose? would like to move to py.test for html5lib, it's compat is such that we don't have to change anything except for requirements-test.txt and tox.ini/.travis.yaml |
| 13:53 | <hemanth> | meow! |
| 14:55 | <gsnedders> | jgraham: argh why did you let #177 go through? it should update the readme and requirements*.txt files too! |
| 16:02 | <jgraham> | gsnedders: Presumably because you didn't review it :p |
| 16:03 | <jgraham> | That said |
| 16:04 | <jgraham> | I think the idea of a changelog is largely dumb. If you care look at the commit history |
| 16:05 | <gsnedders> | I think changelogs are worthwhile for significant changes, or changes that affect a lot of users |
| 16:05 | <gsnedders> | I certainly think they're pointless if you include everything in them |
| 16:05 | <gsnedders> | And commit history is messy when you start having merge commits (bah, GitHub) |
| 16:06 | <gsnedders> | jgraham: can we go back to trying to have a linear history for html5lib? :P |
| 16:06 | <jgraham> | I also note that ordereddict is already in the requirements file for 2.6, so I'm not sure what change you wanted |
| 16:06 | <gsnedders> | it's in requirements-optional-2.6.txt which makes it sound very optional? |
| 16:06 | <gsnedders> | whereas the intention of #177 is that it is now required |
| 16:07 | <jgraham> | I claim it's optional in the sense that if you opt to use 2.6 it's required |
| 16:07 | <gsnedders> | Though I'm now sure we want that. The argument seems to be "this API is broken without this installed". That equally applies html5lib.parse("a", tree="lxml"). |
| 16:07 | <gsnedders> | Should we then say we require lxml, etc.? |
| 16:07 | <gsnedders> | I think the whole argument that it's required on 2.6 is bogus, tbh |
| 16:11 | <jgraham> | Well python's optional dependency handling is pretty terrible. Not that I can think of a system where it's better |
| 16:12 | <gsnedders> | tl;dr: we should either document it as being required or we should not list it as required in setup.py |
| 16:13 | <gsnedders> | jgraham: fyi I'm moving us over to py.test, because I think it makes all our stuff with loading tests from external files way saner |
| 16:13 | <gsnedders> | and makes writing a script to update expected failures sane :P |