1 2011-12-17T00:02:36 <izibi> WHY? :D
2 2011-12-17T00:03:44 <ThomasWaldmann> (23:38) < ThomasWal> i don't think waiting time helps
3 2011-12-17T00:05:58 <izibi> it takes longer until it crashes :D
4 2011-12-17T00:13:10 <ThomasWaldmann> n*2s maybe? :P
5 2011-12-17T00:16:37 <izibi> is that supposed to happen? o.O http://paste.pocoo.org/show/521920/
6 2011-12-17T00:17:45 <ThomasWaldmann> i don't think so, #pypy :)
7 2011-12-17T00:18:54 <izibi> but it's not reproducable
8 2011-12-17T00:19:23 <ThomasWaldmann> well, shouldn't happen I think
9 2011-12-17T00:19:54 <izibi> increasing the file limit (to 40k) didn't help
10 2011-12-17T00:19:57 <izibi> very strange
11 2011-12-17T00:27:51 <izibi> why is there no traceback or something? :(
12 2011-12-17T00:48:02 *** raignarok
13 2011-12-17T00:53:43 *** raignarok
14 2011-12-17T01:57:55 *** raignarok
15 2011-12-17T02:09:35 *** raignarok
16 2011-12-17T02:18:26 *** qxcv
17 2011-12-17T03:09:15 *** raignarok
18 2011-12-17T03:09:56 *** raignarok
19 2011-12-17T03:12:42 *** raignarok
20 2011-12-17T03:32:23 *** raignarok
21 2011-12-17T13:40:19 *** raignarok
22 2011-12-17T14:56:56 *** qxcv
23 2011-12-17T15:57:01 <ThomasWaldmann> eSyr: http://www.google-melange.com/gci/task/view/google/gci2011/7121279 another fast one...
24 2011-12-17T16:46:00 *** raignarok
25 2011-12-17T17:07:26 *** raignarok
26 2011-12-17T17:15:36 <izibi> now i've compiled openssl 0.9.8 myself to use the official pypy build and it still crashes
27 2011-12-17T17:15:57 <izibi> and i was able to reproduce this on 3 different hosts (at least with the arch build)
28 2011-12-17T17:16:24 <ThomasWaldmann> crashes means?
29 2011-12-17T17:18:34 <izibi> http://paste.pocoo.org/show/522108/
30 2011-12-17T17:19:51 <ThomasWaldmann> maybe it is because of apophis :P
31 2011-12-17T17:20:29 <izibi> i could reproduce it on 3 hosts (well, all arch)
32 2011-12-17T17:21:32 <izibi> but i don't know what else could be wrong
33 2011-12-17T17:26:19 <ThomasWaldmann> hmm
34 2011-12-17T17:26:59 <ThomasWaldmann> no debian/ubuntu in reach?
35 2011-12-17T17:28:25 <ThomasWaldmann> while I would accept the work you have done for that task, I would really like that task being done. but once it is closed, it is gone.
36 2011-12-17T17:28:52 <izibi> i could set up a virtual machine
37 2011-12-17T17:29:30 <izibi> but then i'd have to run the other benchmarks again
38 2011-12-17T17:29:55 <izibi> or i could just use a chrott
39 2011-12-17T17:29:58 <izibi> *chroot
40 2011-12-17T17:30:04 <izibi> should be fine for that
41 2011-12-17T17:32:16 <ThomasWaldmann> well, if you think that isn't too much...
42 2011-12-17T17:33:04 <ThomasWaldmann> another option might be documenting the pypy1.7 issues, filing bugs to arch/pypy trackers
43 2011-12-17T17:33:50 <ThomasWaldmann> "does not work" is also a testing outcome, although we expected working :)
44 2011-12-17T17:36:21 <ThomasWaldmann> also you could do some analysis / filing issues for the py.test different outcomes you have already found
45 2011-12-17T17:38:07 <ThomasWaldmann> did you show the speed results to the pypy guys, btw?
46 2011-12-17T17:40:34 <ThomasWaldmann> izibi: please give more precise cpu type on results page. 2.5GHz dual core can mean a lot of different things
47 2011-12-17T17:42:46 <izibi> ThomasWaldmann: nope, i only showed the crash to them. currently trying to find out what's going with gdb. that's fun
48 2011-12-17T17:59:23 <izibi> wtf
49 2011-12-17T17:59:40 <izibi> might be a problem with werkzeugs reloader
50 2011-12-17T17:59:52 <izibi> when i disable it, it doesn't crash
51 2011-12-17T17:59:54 <ThomasWaldmann> -r
52 2011-12-17T18:00:09 <ThomasWaldmann> document that on the results page
53 2011-12-17T18:01:09 <izibi> well now i want to know what's going on there :P
54 2011-12-17T18:06:54 <izibi> and the subprocess segfaults. yeah!
55 2011-12-17T18:07:01 <izibi> http://paste.pocoo.org/show/522128/
56 2011-12-17T18:15:26 <izibi> 18:11 <@exarkun> Maybe you should switch to hosting moin in Twisted. >:P
57 2011-12-17T18:21:04 <izibi> without the reloader it works fine. (well, too many files open but apart from that)
58 2011-12-17T18:25:58 <ThomasWaldmann> izibi: reloader is part of werkzeug, file a bug there :)
59 2011-12-17T18:26:31 <izibi> i guess it's a very stupid combination of everything :D
60 2011-12-17T18:26:40 <ThomasWaldmann> #pocoo and mitsuhiko
61 2011-12-17T18:32:07 <izibi> in the end it's something like moin triggers a bug in pypy that only happens if it's started like werkzeug does :D
62 2011-12-17T18:32:56 <ThomasWaldmann> well, as werkzeug is quite popular...
63 2011-12-17T18:33:28 <izibi> how large is the performance impact of the reloader?
64 2011-12-17T18:33:46 <izibi> so is there a large difference if i run the benchmarks on pypy 1.7 without the reloader?
65 2011-12-17T18:36:07 <ThomasWaldmann> i would suspect it to be about none
66 2011-12-17T18:42:03 <izibi> < mitsuhiko> izibi: yeah, the segfault case is shitty
67 2011-12-17T18:42:25 <izibi> okay... :D
68 2011-12-17T18:48:33 <TheSheep> ThomasWaldmann: what would you say about this kind of look for moin? http://elementaryos.org/docs/human-interface-guidelines
69 2011-12-17T18:49:36 <TheSheep> ThomasWaldmann: I mostly mean cleanly separated content area and navigation
70 2011-12-17T18:52:19 <ThomasWaldmann> left sidebar looks pretty, might be just a matter of fonts and colour choice
71 2011-12-17T18:53:05 <ThomasWaldmann> trail also quite pretty
72 2011-12-17T18:53:54 <ThomasWaldmann> but that trail is location breadcrumbs, not path breadcrumbs
73 2011-12-17T18:54:57 <ThomasWaldmann> no item related stuff there, we need that for moin
74 2011-12-17T18:55:36 <ThomasWaldmann> and it is not fluid
75 2011-12-17T18:56:12 *** raignarok
76 2011-12-17T18:57:21 <ThomasWaldmann> not much space for a logo, if logo height is bigger than font height of "elementary" there, it'll look less great
77 2011-12-17T18:58:32 <ThomasWaldmann> i also think we need more space in sidebars. there's nothing left there to use, all sidebar space taken, but lots of unused screen space.
78 2011-12-17T19:00:15 <izibi> please don't try to make a apple-ish theme :D
79 2011-12-17T19:00:56 <ThomasWaldmann> is that apple-ish?
80 2011-12-17T19:04:04 <izibi> imo yes
81 2011-12-17T19:04:32 <ThomasWaldmann> TheSheep: doesn't look nice with < 960px, worse at < 920px, unusable at 860px
82 2011-12-17T19:07:02 <ThomasWaldmann> the look in general is nice and smooth, though. but looks is not everything.
83 2011-12-17T19:09:17 <izibi> http://paste.pocoo.org/show/522161/ can someone try this on another distribution?
84 2011-12-17T19:09:50 <izibi> then use "ab -n 1000 http://127.0.0.1:8080/Home" to (maybe?) crash it
85 2011-12-17T19:10:56 <ThomasWaldmann> i ran this yesterday
86 2011-12-17T19:12:03 <izibi> on which distribution?
87 2011-12-17T19:12:12 <ThomasWaldmann> ubuntu 11.10
88 2011-12-17T19:15:43 <izibi> let's set up a debian chroot and see what happens there
89 2011-12-17T19:17:28 <TheSheep> ThomasWaldmann: I was asking about the looks
90 2011-12-17T19:18:06 <ThomasWaldmann> colours, fonts, shapes are nice
91 2011-12-17T19:18:35 * ThomasWaldmann is always looking at functionality also :)
92 2011-12-17T19:19:07 <ThomasWaldmann> hmm, shall we make a branch for the new theme stuff?
93 2011-12-17T19:20:51 <TheSheep> ThomasWaldmann: if you ask me, then don't bother until you have someone who can take the lead
94 2011-12-17T19:21:17 <TheSheep> ThomasWaldmann: I hope to do it at some time, but right now it's rather hectic here
95 2011-12-17T19:30:29 <ThomasWaldmann> ok
96 2011-12-17T19:35:22 <izibi> < Da_Blitz> izibi: i have it exit after a couple of reqeusts as well on debian with pypy 1.7 from the website
97 2011-12-17T19:35:34 <izibi> looks like it's not specific to arch
98 2011-12-17T19:35:52 <izibi> why doesn't it happen on ubuntuu? do they apply some weird patchs to something?
99 2011-12-17T19:37:51 <izibi> why don't pocoo and pypy take part in the GCI? :D
100 2011-12-17T19:38:06 <izibi> plenty of work for me :D
101 2011-12-17T19:39:09 <ThomasWaldmann> btw, i don't think we need "local" and "network"
102 2011-12-17T19:39:13 *** raignarok
103 2011-12-17T19:39:20 <TheSheep> I thik pypy takes part in gsoc
104 2011-12-17T19:39:33 <ThomasWaldmann> you can always test locally, the network performance will be same minus latency
105 2011-12-17T19:43:45 <ThomasWaldmann> izibi: btw, you should also specify what is on your "Home" page as rendering speed depends on that.
106 2011-12-17T19:44:16 <izibi> just some lorem ipsum ;)
107 2011-12-17T19:47:17 <izibi> does somebody know a nice tool for rendering graphs?
108 2011-12-17T19:47:39 <izibi> or charts or however you want to call it ;)
109 2011-12-17T20:21:39 <ThomasWaldmann> the numbers are enough
110 2011-12-17T20:21:54 <ThomasWaldmann> eSyr: want to help trying out transifex?
111 2011-12-17T20:22:51 <izibi> but charts are awesome: http://moinmo.in/MoinOnPyPyTesting
112 2011-12-17T20:24:32 <ThomasWaldmann> hehe
113 2011-12-17T20:24:48 <ThomasWaldmann> show to the pypy guys, they usually want the charts the other way round
114 2011-12-17T20:32:09 *** raignarok
115 2011-12-17T20:32:25 *** raignarok
116 2011-12-17T20:46:47 <CIA-59> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 1161:bec6f9485403 2.0/MoinMoin/translations/MoinMoin.pot: update MoinMoin.pot
117 2011-12-17T21:21:38 <ThomasWaldmann> https://www.transifex.net/projects/p/MoinMoin/
118 2011-12-17T21:29:27 <izibi> http://www.google-melange.com/gci/task/view/google/gci2011/7185227 marked task as complete
119 2011-12-17T21:29:56 <izibi> and now i'll look at werkzeug/pypy again
120 2011-12-17T21:32:46 <izibi> ThomasWaldmann: btw do you use a 32 or 64 bit ubuntu?
121 2011-12-17T21:36:39 <ThomasWaldmann> 32 (with pae)
122 2011-12-17T21:37:44 <izibi> that could be a reason why it works for you
123 2011-12-17T21:39:03 * ThomasWaldmann likes working stuff :)
124 2011-12-17T21:40:50 <izibi> ok, chrome uses multiple processes, so it can use more than 4 gig on ram even on a 32bit system :D
125 2011-12-17T21:41:05 <ThomasWaldmann> izibi: btw, "moin moin" also uses 8080
126 2011-12-17T21:41:58 <izibi> i hope that won't make a difference in performance :D
127 2011-12-17T21:43:20 <ThomasWaldmann> izibi: google will be proud of you :D
128 2011-12-17T21:43:55 <izibi> because i used their chart api?
129 2011-12-17T21:44:03 <ThomasWaldmann> and go
130 2011-12-17T21:44:15 <ThomasWaldmann> did you do "manual testing"?
131 2011-12-17T21:44:55 <izibi> yes, but i didn't find any problems
132 2011-12-17T21:45:06 <ThomasWaldmann> ok, put a note on the wiki page
133 2011-12-17T21:45:24 <ThomasWaldmann> also link to issues you created on trackers
134 2011-12-17T21:45:47 <izibi> only thing was that the css didn't load properly at the beginning. but that got resolved by pressing F5
135 2011-12-17T21:46:07 <izibi> didn't find any failing requests in the log though
136 2011-12-17T21:47:07 <izibi> i'm still looking at that issue, so no report yet
137 2011-12-17T21:47:58 <izibi> you can leave the task pending if you want to, i won't do something for gci in that time anyway ;)
138 2011-12-17T21:58:06 <izibi> added section on manual testing
139 2011-12-17T21:58:23 <izibi> 21:47 < rguillebert> I have the same bug
140 2011-12-17T21:58:23 <izibi> 21:47 < rguillebert> on 32bits
141 2011-12-17T21:58:24 <izibi> lol
142 2011-12-17T21:58:31 <izibi> then that's not the problem :D
143 2011-12-17T22:03:03 <ThomasWaldmann> you found that reloader issue, for example
144 2011-12-17T22:04:17 <izibi> mitsuhiko's reaction looked like he knows that this sucks :D
145 2011-12-17T22:05:30 <ThomasWaldmann> well, you could check werkzeug's issue tracker
146 2011-12-17T22:06:55 <ThomasWaldmann> or in pypy's?
147 2011-12-17T22:08:57 <izibi> well the proper handling of the exit code is a problem in werkzeug
148 2011-12-17T22:09:11 <izibi> but the segfault might be in pypy, that's what i'm currently trying to find out
149 2011-12-17T22:09:22 <izibi> as it worked on pypy 1.6
150 2011-12-17T22:10:08 <ThomasWaldmann> btw, there is also an easy to fix issue (maybe) in moin
151 2011-12-17T22:10:56 <ThomasWaldmann> http://moinmo.in/MoinOnPyPyTesting?action=AttachFile&do=view&target=py.test-pypy1.7.txt#CA-55e41a4bb7a849e19fac3826b9b212c93a04c612_342
152 2011-12-17T22:11:06 <ThomasWaldmann> you could try if a .close() helps there
153 2011-12-17T22:11:19 <ThomasWaldmann> or "with"
154 2011-12-17T22:11:43 <izibi> which one?
155 2011-12-17T22:14:00 <ThomasWaldmann> see that long url
156 2011-12-17T22:15:29 <izibi> after line 348?
157 2011-12-17T22:15:34 <izibi> or where to close it?
158 2011-12-17T22:16:28 <ThomasWaldmann> after using the file :) and before reopening it
159 2011-12-17T22:17:13 <ThomasWaldmann> that paramparser issue is also strange and fails on 1.6 AND 1.7 (unlike most other failures on 1.6)
160 2011-12-17T22:17:50 <ThomasWaldmann> so check if that is easily fixed and if not, file an issue about it
161 2011-12-17T22:23:34 <izibi> tests are running so slow on pypy :D
162 2011-12-17T22:24:10 <izibi> would be interesting to know why it takes longer on 1.7 than on 1.6
163 2011-12-17T22:24:29 <izibi> is this because fewer tests fail which means more code gets executed?
164 2011-12-17T22:24:43 <izibi> or did something in pypy get worse?
165 2011-12-17T22:25:30 *** RogerHaase
166 2011-12-17T22:28:05 <izibi> is the with statement available in 2.6 by default or do i have to import it?
167 2011-12-17T22:28:15 *** raignarok
168 2011-12-17T22:28:28 <izibi> but this fixes the test
169 2011-12-17T22:28:57 <ThomasWaldmann> builtin
170 2011-12-17T22:29:59 <izibi> http://paste.pocoo.org/show/522254/
171 2011-12-17T22:31:04 <ThomasWaldmann> syntax error?
172 2011-12-17T22:31:38 <izibi> huh?
173 2011-12-17T22:32:27 <ThomasWaldmann> with open(...) as f:
174 2011-12-17T22:33:16 <izibi> but it worked :D
175 2011-12-17T22:33:44 <izibi> no, it didn't
176 2011-12-17T22:33:49 <izibi> omg self-fail :D
177 2011-12-17T22:34:09 <izibi> when you get an error at the top instead of an failure on the bottom :D
178 2011-12-17T22:36:51 *** raignarok
179 2011-12-17T22:37:13 <izibi> but now it really works :D
180 2011-12-17T22:37:31 <ThomasWaldmann> so that fixed the issue on all pythons?
181 2011-12-17T22:38:55 <izibi> it was only broken on 1.7
182 2011-12-17T22:40:18 <izibi> i'll commit it
183 2011-12-17T22:43:24 <izibi> https://bitbucket.org/julianbrost/moin-2.0/changeset/b826684567f7
184 2011-12-17T22:51:49 <ThomasWaldmann> ok, any idea about the other one?
185 2011-12-17T22:54:13 <izibi> let's have a look
186 2011-12-17T23:00:01 <izibi> http://paste.pocoo.org/show/522269/ so that's what happens on cpython 2.7
187 2011-12-17T23:01:18 <izibi> is that really what is supposed to happen? :D
188 2011-12-17T23:03:24 <ThomasWaldmann> ehrm
189 2011-12-17T23:04:44 <izibi> http://paste.pocoo.org/show/522276/
190 2011-12-17T23:06:54 <izibi> i don't know if that's really something you want to test :D
191 2011-12-17T23:13:33 *** RogerHaase
192 2011-12-17T23:18:45 <ThomasWaldmann> hmm, no idea, strange stuff
193 2011-12-17T23:19:07 <izibi> http://hg.moinmo.in/moin/2.0-dev/rev/2557 there this test got introduced ;)
194 2011-12-17T23:24:30 <ThomasWaldmann> Changed in version 2.6: For 3.0 forward-compatibility, im_func is also available as __func__, and im_self as __self__.
195 2011-12-17T23:24:43 <ThomasWaldmann> can you try __func__?
196 2011-12-17T23:26:44 <izibi> same as with im_func
197 2011-12-17T23:26:48 <izibi> works in pypy
198 2011-12-17T23:26:54 <izibi> AttributeError in cpython
199 2011-12-17T23:29:07 <izibi> added a bug report for werkzeug: https://github.com/mitsuhiko/werkzeug/issues/144
200 2011-12-17T23:39:25 <ThomasWaldmann> http://kbyanc.blogspot.com/2007/07/python-more-generic-getargspec.html
201 2011-12-17T23:45:02 <izibi> i don't really know/understand what that function should do. i guess i'll create a bug for this :P