IRC log of #schooltool for Friday, 2006-01-27

*** Theuni has joined #schooltool00:17
Theunihowdi00:17
TheuniI don't suppose anybody to be alive right now, hmm?00:17
* Theuni waves to Aiste 00:18
Theuniok then. i'll just jump into my bed too ...00:23
Theunii'll try again tomorrow00:23
*** Theuni has quit IRC00:23
*** jinty has joined #schooltool00:37
*** Travis_ has joined #schooltool00:54
*** th1a has joined #schooltool01:20
*** jinty has quit IRC01:24
*** wrobel has quit IRC01:27
*** Travis_ has left #schooltool01:43
*** Travis_ has joined #schooltool04:02
*** Travis_ has left #schooltool04:04
*** jinty has joined #schooltool04:55
*** jinty has quit IRC07:59
*** wrobel has joined #schooltool09:35
*** mgedmin has joined #schooltool12:51
*** srichter has joined #schooltool12:53
*** alga has joined #SchoolTool13:20
*** strichter has joined #schooltool13:31
*** jinty has joined #schooltool13:36
*** ignas has joined #schooltool14:04
*** srichter has quit IRC14:04
*** jinty has quit IRC14:16
*** jinty has joined #schooltool14:21
*** jinty has quit IRC14:38
*** mgedmin has quit IRC14:51
*** strichter is now known as srichter15:02
*** mgedmin has joined #schooltool15:08
povbot/svn/commits: * ignas committed revision 5695:15:09
povbot/svn/commits: Make SectionAttendance use the filtering provided by OOBTree.15:09
* mgedmin cheers ignas15:13
*** mgedmin has quit IRC15:14
*** mgedmin has joined #schooltool16:02
ignasfull sample data ftest started swapping again :/16:10
mgedminignas, shut down X and use links for development ;-)16:13
ignasmgedmin, won't help i am afraid,  only_last_n_days = 5 is a lifesaver though ;)16:13
*** srichter has quit IRC16:22
povbot/svn/commits: * ignas committed revision 5696:17:43
povbot/svn/commits: Sample data for day attendance.17:43
mgedminyay!17:44
mgedminis there a way to pretty-print libxml2 nodes?17:44
mgedminI've been thinking...18:03
mgedminsample data plugins are utilities18:03
mgedminutilities are created once and stored forever18:04
mgedminso if a utility wants to keep state in attributes of `self`, it could end badly18:04
povbot/svn/commits: * mg committed revision 5697:18:08
povbot/svn/commits: List of absences per term.18:08
povbot/svn/commits: The checkin ought to have been split into smaller ones, if I had more time.18:08
povbot/svn/commits: One noteworthy bit is the way student-attendance.txt extracts a single <table> element from the HTML and pretty-prints it.  This could be a useful addition to the schooltool.testing.analyze module.18:08
mgedminjikes, test.py -f --level 2 eats 450 mb RSS now and is still growing18:32
mgedmin520 megs now18:34
mgedmin600 megs18:37
mgedmin700 megs18:39
mgedmingone18:39
mgedmintime ./test.py -pvC1 -s src/schooltool/ -f --level 218:39
mgedminreal    18m21.954s18:39
mgedminuser    14m39.909s18:39
mgedminsys     0m12.675s18:39
ignasmgedmin, how is the object quantity different after the addition of day attendance ?18:41
mgedminno idea18:42
mgedminfunctional tests use an in-memory db18:42
mgedminnow I've changed the sample attendance code to store lists of person names instead of person object references18:42
mgedminlet's see if that changes memory usage at all18:42
mgedmin@ 2 minutes of cpu time: 250 megs18:44
mgedmin@ 4 minutes of cpu time: 285 megs18:47
mgedmin@ 6 minutes of cpu time: 360 megs18:49
mgedmin@ 8 minutes of cpu time: 455 megs18:52
mgedmin@10 minutes of cpu time: 540 megs18:55
mgedmin@12 minutes of cpu time: 660 megs18:57
mgedminand done18:57
mgedminmg@pitonas:~/src/schooltool $ time ./test.py -pvC1 -s src/schooltool/ -f --level 2 . sampledata/ftests/sampledata.txt18:57
mgedminreal    15m15.855s18:57
mgedminuser    11m34.066s18:57
mgedminsys     0m11.973s18:57
mgedminthe question is: was the memory size decrease because of my change, or because I excluded all other functional tests?18:58
mgedminignas, want a diff?19:00
mgedminhere: http://schooltool.pastebin.com/52587119:00
mgedminit breaks unit tests that substitute strings for person objects19:00
th1aTalking to yourself, mgedmin?19:03
mgedminlet's try a live run!19:06
ignasth1a, nop19:06
ignasth1a, talking to me19:06
* ignas was making food19:06
th1aYes, I saw.19:07
th1aignas:  I see it is warming up in Vilnius.19:07
ignasyep19:07
* mgedmin wonders if it would be possible to leverage modern 3D accelerated graphics cards to speed up sample data generation19:07
ignaswe would need to adapt our data structures to ones used in 3D graphics19:08
ignaswhich would require too many changes19:08
ignasand we should spend the effort in just plainly optimizing the data structures in zope3 instead ;(19:09
ignas;) i mean19:09
mgedminbtw those numbers were VIRT; RSS was about 20 megs lower19:10
mgedminlive instance has a much larger difference between RSS and VIRT19:10
mgedminabout 90 megs19:10
mgedminhmm, buildbot is impatient again :)19:16
mgedminRSS for live sample data is steady at 172 megs (VIRT at 258 megs)19:16
mgedminthose transaction.commit()s help when you're using a disk-based ZODB backend19:17
povbot/svn/commits: * mg committed revision 5698:19:20
povbot/svn/commits: Move the sampledata functional test to level 3, so that buildbot won't try to run it so often.19:20
povbot/svn/commits: This functional test currently requires 700 megs of free memory, and about 20 minutes on a 1.6 GHz Pentium M.  If your machine does not have that much RAM, the ftest starts swapping and becomes unbelievably slow (buildbot kills it after waiting for 1 hour).19:20
ignasmgedmin, maybe setting the day limit to 5 would be a better idea ?19:26
ignasso we would at least test whether it actually works19:26
mgedminit's done in 11 minutes (and at 186 megs RSS)19:27
mgedminignas, good idea19:27
mgedminData.fs is 400 megs19:27
mgedmin300 megs are in current objects19:30
ignasmgedmin, what data should we log for attendance ?19:30
mgedminignas, you wanted object statistics?19:31
mgedminhttp://schooltool.pastebin.com/52591819:31
mgedminre: loging -- why don't you ask th1a?19:31
ignasth1a: would "date, manager_id, student_id, action" be enough19:32
ignasor we need actual explanations19:32
th1aOh... good point...19:32
th1aDon't include the explanations at this point.19:32
* mgedmin discovers that is is not possible to pack the schooltool Data.fs, because ++etc++process sidesteps schooltool's authentication19:33
ignasmgedmin, as far as i can see the amount of space required by the activities/ processes increased considerably19:34
mgedminyes19:34
ignasthat's because i am leaving most of the daily abscences unexplained, and each daily abscence generates 4-5 unexplained section abscences ...19:36
th1aGenerally the business process of schools will probably be to resolve all absences after a reasonable period of time, say two weeks.19:37
th1aSo we'll probably be adding that to the workflow somehow.19:37
th1aBut in the meantime I wouldn't assume they're mostly unresolved in the sample data.19:38
th1aIf that's what's ballooning the size of things.19:38
ignasactually - sectio abscences generated by day abscences are allways unresolved ..19:38
th1aWell, that's not good.19:38
th1aThey need to be linked somehow, I guess.19:38
*** tiredbones has left #schooltool19:42
povbot/svn/commits: * ignas committed revision 5699:20:08
povbot/svn/commits: Added an optimization to the generation of sample attendance data.  Moved day_absences into a parameter (temporary fix).20:08
*** mgedmin has quit IRC20:13
*** alga has quit IRC20:26
*** Aiste has quit IRC20:28
*** tiredbones has joined #schooltool20:47
*** ignas has quit IRC20:58
*** erchache has joined #schooltool21:23
erchacheth1a: are you here?21:24
th1aerchache:  Yes.21:24
erchacheDo you investigate how to translate non-archetypes objects with linguaplone?21:25
erchacheth1a21:25
th1aStephan and I spent a while trying to figure it out.21:26
th1aAnd had no luck.21:26
erchachearrghh21:26
erchachei need to work on that....dont worry21:26
erchachei was working on my backup system...i finished....now i have more time to spend on schooltool21:27
th1aerchache:  I'm starting to work on something you needed.21:27
th1aA way to script the setup of your entire school.21:28
erchachefor?21:28
th1aSo you can easily redo the whole thing if some kind of change needs to be done.21:28
th1aAlthough this will only work when you're initially setting up the site,  I guess.21:28
erchacheahhh ok...a temporal solution to recursive events21:29
erchacheok21:29
erchachegreat21:29
th1aOr that's what it is designed for.21:29
th1aWell, no21:29
erchacheim fighting with one of my bosses to get a payed full-time programer student to work on schooltool21:29
th1aIsn't there some kind of issue with your timetables that can't really be changed at this point?21:29
th1aerchache:  That would be fantastic.21:30
erchachewell.....recursive event greater than a month21:30
th1aIf there is anything I could do to help make that happen (write a letter or something).21:30
erchachebecause on this university we planned classes 6 month before21:30
erchachewe work with semester...21:30
erchacheand when we try to put into schooltool.....breaks on start point....enters in a infinite loop and doesnt have a unique start poiint21:31
erchachedont worry.....next week i get time to plan my work and my realistic problems...21:32
th1aerchache:  I'm confused about whether or not that bug has been fixed in svn.21:32
erchachemgedmin was working on that....but now i dont remember....sorry....my brain is like hot butter now21:33
erchache:-S21:33
erchachewell i left....see all next monday :-D21:34
erchachebye21:34
*** erchache has quit IRC21:34
*** th1a|inux has quit IRC22:34

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!