IRC Logs for #circuits-dev Monday, 2013-02-18

*** Osso has joined #circuits-dev01:05
prologichey Osso01:10
prologichow's things?01:10
prologicI didn't release on Sunday (scheduled)01:10
prologicbecause ofa few reasona01:10
Ossocrazy busy over here :)01:10
prologicmainly some failing unit tests for windows01:10
prologicsome unicode issues still01:10
prologicand ipv6 tests for me are failing01:11
prologicyeah crazy busy over here too :)01:11
Ossooh no!01:15
Ossois the unicode failing some tests ?01:16
Ossoyeah it is failing here01:18
Ossonow it's goes into the evil spiral of doom01:20
Ossowhen I run tests01:20
Ossodefinitely not ready to realise :D01:20
prologicno we're not :)01:25
prologicthe unicode issue(s) are centered around the fact taht some platforms/systems have different default encoding settings01:25
prologicwe need to ensure our code is paltform independent in that case I think01:25
prologicand work no matter what the system default is01:25
prologicif that makes sense?01:25
prologicas for the ipv6 tests failing here for me on my desktop - I have no idea why01:25
prologicI keep getting "network errors"01:25
prologicMark Young is still around and emailed me today - so hopefully he can help out with the remaining 6 windows tests taht are failing01:26
prologicalso BSD01:26
*** koobs has joined #circuits-dev01:27
prologicalso Osso01:27
koobsBoom goes the dynamite01:27
prologickoobs will kindly donate a BSD shell to do some testsing against01:27
prologicthank you koobs :)01:27
Ossothe unicode test is failing here01:27
prologicas for buildbot - I'm not sure if Jenkins/shiningpands supports that01:27
koobsHave two official core python buildbots running on this machine, happy to run a dedicated circuits one if you guys can arrange a master01:27
Ossowe encoding some string that is already unicode01:27
prologicI'd have to look into how to get a BSD slave working with shiningpanda01:27
koobsprologic; as long as it doesnt require i run java ;)01:28
prologicI'll get back to you on that01:28
prologicI *think* shiningpanda can support remote slave systems01:28
prologicOsso, yeah - I'm about as good at unicide in Python 2 as the next guy I'm afraid :/01:29
prologicHey - Did you guys see my first blog post on Planet Python?01:29
OssoI can fix the unicode tests I just need to manage to find some free time01:29
OssoI saw it prologic01:29
OssoI was like I know this name!01:29
prologicThat's okay, we'll delay the release until we sort these stories out01:29
prologicOsso, :)01:30
prologicI finally took up many years of suggestions "to write/maintain a blog"01:30
koobsHard habit to get into isnt it :)01:43
prologicyeah :)02:07
prologicOsso, what have you concluded if anything re the memory leak story?03:09
Ossonope sorry I did not have time :(04:10
OssoI'll have at least an hour today04:11
Ossobut I'll fix the unicode first04:11
Ossoand hear goes my free our I was supposed to have ~05:49
*** anunnaki has quit IRC06:47
Ossoprologic: it's not that catastrophic07:06
Ossoby default you get bytes07:06
Ossoin py207:06
Ossothen you decode with the encoding you specify07:07
Ossoso the fact that you get the ascii mostly means we are decoding twice07:07
prologicI was just going off what Michael said :)07:07
prologicI'll see if I can create a set of unit tests for testing against memory leaks07:07
prologicso that the test will fail if we introduce a memory leak into the core07:08
Ossothe problem is that py3 gives you a unicode string07:08
Ossoso it is horrible to be get both py2 and py3 working :(07:08
prologiccircuits.six module won't help?07:08
prologicI integrated that into our codebase so out py3 support would not be so painful nor ugly in terms of no. of loc07:09
prologichmmm 2am07:09
prologicI need to go back to bed :)07:09
prologicjust wanted to blog about the python trademark issue07:10
prologicand reorganize the stories holding us up from 2.1.0 :)07:10
Ossoyeah there is this u() function but07:10
Ossoit's encoding with "unicode_escape" ?07:11
prologicstr(foo, "unicode_escape")07:13
prologicis equivalent to07:13
Ossoyeah though07:13
Ossoit was for replacing the u""" syntax07:13
Ossowe have to use six.binary_type07:15
prologicyes I think so07:15
prologicI've used that in a few places already in the py3 port07:15
Ossoyou have to admit, it is not very nice07:16
prologicat least with sockets (I/O) this is easy07:16
prologicbecause socket streams must be bytes07:16
prologicwhat isn't?07:16
Ossochecking if it is an instance of bytes07:17
prologicahh yeah07:17
prologicwhilst I agree07:17
prologicprice we pay for a crappy first cut impl. of unicode support in Python 2.x :)07:17
Ossoit's like admitting it is out of control and we don't know what's coming through :D07:17
prologichopefully in a few years we can drop 2.x support too :) haha07:17
prologicI agree07:17
*** Osso has quit IRC07:45
mehereAnother attempt to fix that nasty str/bytes/unicode stuff. BTW Shining Panda seems to run out of credit...09:22
mehereThe test seems to be better now. Shining Panda shows an error that I actually did expect. Strange thing is that it doesn't occur locally in my environment.09:30
mehereGot it, trying to fix...09:33
mehereDid you know that we currently don't have any charset handling at all on the server side?09:56
mehereJust noticed that decoding is done by cgi.FieldStorage. Well, it works here for both py26 and py32. I have removed the TextIOWrapper (seems to be a common problem, they have added explicit removal of that wrapper in the py32 lib). Have to wait until Shining Panda reports something again.11:23
*** anunnaki has joined #circuits-dev11:53
prologicmehere, thanks, I'll top up my account today12:57
*** anunnaki has quit IRC13:13
*** anunnaki has joined #circuits-dev14:28
*** anunnaki has quit IRC14:34
*** anunnaki has joined #circuits-dev14:52
*** anunnaki has quit IRC21:04
*** anunnaki has joined #circuits-dev23:17

Generated by 2.11.0 by Marius Gedminas - find it at!