IRC log of #schooltool for Monday, 2005-09-12

*** mgedmin has quit IRC00:08
*** munkee has quit IRC05:54
*** munkee has joined #schooltool05:55
*** tvon has quit IRC07:11
*** tvon has joined #schooltool09:41
*** mgedmin has joined #schooltool12:11
*** mgedmin has quit IRC12:11
*** mgedmin has joined #schooltool12:15
*** ignas has joined #schooltool12:21
*** jinty has joined #schooltool12:58
*** thisfred has joined #schooltool13:02
*** Ricey has quit IRC13:02
*** Ricey has joined #schooltool13:05
*** Ricey_ has joined #schooltool13:14
*** Ricey has quit IRC13:15
*** Aiste has joined #schooltool13:56
*** Aiste has quit IRC13:58
*** Aiste has joined #schooltool13:59
*** Aiste has joined #schooltool14:00
povbot/svn/commits: * ignas committed revision 4947:14:05
povbot/svn/commits: Added hasPermissions function for batch permission checking.14:05
*** didymo has joined #schooltool14:12
*** strichter has joined #schooltool14:20
*** jinty has quit IRC14:20
*** strichter has quit IRC14:25
*** srichter has quit IRC14:34
*** didymo has quit IRC14:57
*** appyR has joined #schooltool15:03
*** srichter has joined #schooltool15:31
*** strichter has joined #schooltool15:32
*** bskahan has joined #schooltool15:52
*** tiredbones has quit IRC16:16
th1a_Hello, all16:30
srichtergood morning16:30
th1a_Is everyone here and awake?16:31
srichterme ;-)16:31
th1a_ignas & medmin?16:32
*** th1a_ is now known as th1a16:33
* srichter cracks up...16:34
th1aWe actually have a lot of important stuff to discuss today...16:40
bskahanhello everyone16:42
th1aWell, we can get started, I guess.16:44
th1abskahan:  How is the server coming?16:45
bskahanI ran into a problem creating the demo chroot for schooltool16:45
bskahanit didn't work the easy way, so now I have to do it the hard way16:45
bskahanubuntu and debian debootsrap both failed for assorted reasons16:45
* th1a wishes we could ask mgedmin how they did it before.16:46
bskahanI'd really rather not run a public demo outside of a chroot16:46
th1aI'm more worried about getting the Plone site up first.16:46
bskahanok, that can be up quickly16:47
bskahanits allready setup and ready to run, I'll install all the packages today and start moving content tomorrow16:47
th1aActually, I'd like to poke around with the site before we start moving the content over.16:48
th1aI think PloneSoftwareCenter and PloneHelpCenter should change the structure of the site quite a bit.16:49
th1aBut I won't really understand it until I get to poke around with them a bit.16:49
srichterwhat will PloneSoftwareCenter and HelpCenter give us?16:50
th1aI haven't been able to get PSC running on my laptop instance.16:50
th1asrichter:  They're basically the products that manage releases and documentation on
th1aThe way we publish releases right now is hopelessly haphazard.16:51
* srichter wonders whether we need this16:51
srichterI am a fan of simple :-)16:51
srichterI see16:51
th1aIt is more a matter of having a structure.16:51
bskahanPSC won't get us alot until we have third party packages, but the PEP forms are a good start for proposals16:51
srichterbut truthfully, I hate the hoops I have to jump through for Zope 3 and wish I could just upload a file16:52
bskahanthe release part of PSC is geared to "collections" of software16:52
th1aWell, right now I don't even handle SchoolBell and SchoolTool exactly the same way, so we've got a lot of room for improvement.16:53
srichterI see16:53
srichterI open to see how it looks16:53
th1aIs Plone running on the server now?  What is the URL?16:54
bskahanits not started at the moment16:54
bskahanadd ""  to your /etc/hosts16:55
th1aWill be the Plone site?16:55
mgedminhi all16:56
mgedminsorry, I've been detained a bit16:56
mgedminjust got back to the office16:56
bskahanth1a: yes16:56
th1amgedmin: Detained by the Belorussian secret service?16:57
bskahanI have to run unfortunately, I'll ping you this afternoon when they're back online16:57
th1aOy.  This is why you have to treat every meeting like it is important, even if it isn't.16:57
mgedmindemo server: all is described16:58
mgedminalthough you probably do not want to use User Mode Linux16:58
mgedminSteveA once told me that some Canonical sysadmin claimed UML is not safe16:58
th1amgedmin:  Oh yeah.  I forgot you guys used UML.16:58
th1aUML is a lot more work, isn't it?16:59
bskahanmgedmin: any reason a chroot wouldn't be sufficient?16:59
th1aSeems to me that chroot would be fine.17:00
mgedminit's up to the admin17:00
mgedminwe had the routing set up so that the UML couldn't make outgoing connections17:00
mgedminso that even if someone found a hole in ST and hacked it, they wouldn't get a free spam proxy or something17:00
bskahanI'll look over the UML documentation, I hadn't considered doing it that way17:01
mgedminchroot doesn't give you that17:01
mgedminXen is another option17:01
* bskahan nods 17:01
mgedminbut then again -- if we do not trust ST enough to run it on our servers17:01
mgedminhow can we ask our users to run it?17:02
th1aI certainly don't see any reason to think it is more hackable than say, Plone.17:02
srichterI agree17:02
srichterwe are not running external OS-level scripts from within ST?17:02
bskahanI think a public demo server is a different case than a project install of schooltool17:02
mgedminhmm, right -- the demo server was not a single ST instance17:03
SteveAthe main issue was schooltool's use of xmllib on outside input on its REST interface.17:03
SteveAum, libxml2, i mean17:03
SteveAit's had buffer overruns before17:04
mgedminit was a small web app that could create new demo instances with separate Unix user accounts etc.17:04
mgedminso it was running some scripts as root inside the UML17:04
th1aWe aren't worrying about resurrecting the full demo system at this point.17:05
bskahanjust a single demo instance17:05
th1aProbably later.17:05
srichtermaybe we should shut off REST for the demo17:05
bskahansrichter: +117:05
ignassrichter, what is the point of shutting down REST ?17:05
th1aDoes that protect us from libxml bugs?17:06
mgedminignas, immunity to libxml2 bugs17:06
bskahanignas: it doesn't seem like that's something people will get alot of utility from in a demo environment17:06
mgedminth1a, yes, I think so17:06
mgedminyou can't upload arbitrary XML files via web views17:06
th1aOK.  We'll do that & chroot is fine for now.17:06
mgedmin... righ?17:06
* bskahan nods17:06
th1aLet's move on.17:06
bskahanok, sorry I really have to go now17:07
bskahanI'll read over the logs and reply this afternoon17:07
th1abskahan:  OK.  You and srichter need to have a serious meeting soon.17:07
* bskahan nods17:07
srichterany time today, when I am online17:07
bskahansrichter: ok, I'll be back this afternoon17:08
*** bskahan has quit IRC17:08
th1aHow did things go last week at POV?17:08
th1aHow is optimization coming?17:09
mgedminmy one day release-our-product task ate all of my week17:09
mgedminignas and alga spent the week optimizing things17:09
ignasalga finished optimizing recurrent events17:09
ignasi have made normal events a bit faster17:10
ignasand now i am working on ACL view17:10
th1aDoes the ACL view do batching?17:10
ignasthe things left are - yearly/monthly/weekly views17:10
ignasth1a, well it does. not that i does them properly ATM17:10
ignasthat's what i am working on17:10
th1aI'm open to whatever kinds of changes will make that view manageable.17:11
th1aSo let me know if  you have any creative ideas.17:11
srichterwhat's the problem?17:11
ignasth1a, well me and alga managed to sped it up like 80 times17:12
ignasor so17:12
ignasso it kind of makes it bearable17:12
th1aWell, that's good.17:12
th1amgedmin:  When are you going to start on the merge?17:14
mgedminas soon as possible17:15
mgedminthat probably means when ignas becomes available to help me17:15
mgedminuntil then I'm pair programming with him17:15
th1aPair programming on optimization?17:16
th1aOK, so that will be a few days at least?17:16
mgedminI'm afraid to say17:16
mgedminI'd like to use this opportunity and ask a question17:17
mgedminis anyone aware of any changes to the current trunk that aren't yet backported to srichter's branch?17:17
th1aI'm sure srichter would prefer if you two work on the merge first, since that will speed up his payment for that work.17:17
mgedminI am unhappy that ignas and alga started optimization work before the merge17:18
th1asrichter ported changes a few weeks ago, I think.17:18
mgedminnow we have three branches that somehow have to be merged instead of two :/17:18
th1aThis is inherently messy.17:19
th1aWe just have to plow through.17:19
ignasth1a, what is the deadline for optimizations ?17:19
mgedminthanks for the encouragement17:19
th1aignas:  I don't know that there is a specific deadline for the optmizations.17:20
ignasi see, well i am pondering whether i should finish all the optimizations17:21
ignasor just do the ACL view17:21
ignasand go help mgedmin17:21
th1aI'd say finish the ACL view, if you've started it.  Then you and mgedmin should do the merge.17:21
mgedminfine by me17:21
srichterOnly the optimization stuff needs to be merged17:22
srichtermgedmin: I am keeping track of all changes to the trunk17:22
srichterI can give you the revs that still need merging17:22
mgedminsrichter, THANKS!!!17:22
th1aWell, at least we're getting somewhere.17:23
th1aMoving on... for a number of reasons, it is looking like it will be best for srichter and bskahan to collaborate on the "pluggable UI" work.17:24
*** tiredbones has joined #schooltool17:24
mgedminhow about this plan of action: switch trunk and srichter's branch ASAP17:25
mgedminthen work on reviewing / merging stuff into the new trunk17:25
mgedminwhile other people also work on the new trunk17:25
mgedminand try not to step on each other's toes17:25
srichteryeah, I think this would be good17:25
srichterwe really need exposure of the refactor stuff to the developers17:26
srichterso that we can work out problems ASAP17:26
th1aI'm not sure what you mean by "while other people also work on the new trunk."17:27
mgedminpluggable UI work, for example17:28
srichterbasically we do not want to stall development due to the lack of the merger17:29
th1aThe merger will happen soon, so I don't think that will be a problem.17:29
th1aI would like to stick with the pattern of having work done in branches and then merged into the trunk.17:30
th1aIt isn't as smooth in svn as it is in bazaar, but I think it is the way to go.17:31
ignasth1a, branches are increasing the cost of any refactoring, code improvement efforts17:31
mgedminth1a, it'd work if we're careful17:31
ignasand it oposes the continuous integration mantra :/17:32
* srichter is not a big fan of branches either, but will do whatever th1a says is good for him :-)17:32
* th1a is more or less repeating what SteveA says is good for him.17:32
th1aBut it makes sense to me.17:32
th1aPerhaps we are making our branches too big.17:33
mgedminSteveA has tools that enable this17:33
mgedminwe don't17:33
th1aThe problem is that I can't really play the role of bdfl on a code level.17:33
mgedminwhile don't have the time/authority17:34
th1amgedmin, you mean bazaar, or other tools?17:34
mgedminbazaar, pqm, infrastructure for code reviews, established development procedures -- all of it17:34
mgedminwhen you do extensive refactoring on one branch, merging another branch becomes painful with subversion17:35
mgedminactually, I don't know whether bazaar makes it easier17:35
th1aOK.  I'll think about it some more.17:35
mgedminbut bazaar does support repeated merges, while you have to explicitly keep track of revision numbers in SVN if you want to do that17:36
srichterwhat is bazaar?17:36
mgedminSteveA suggested that we try out bazaar-ng somewhere in October17:36
mgedminsrichter, do you know GNU arch?17:36
ignassrichter, bazaar is an RCS17:36
srichterdoes it support Windows?17:36
SteveAbaz2 (known as bazaar-ng) does17:37
srichterbecause that's key17:37
SteveAit is written in python17:37
srichterit's a requirement steve does not have17:37
mgedminthat's one of the reasons why we chose SVN rather than arch for schooltool 2 years ago17:37
*** jinty has joined #schooltool17:37
mgedminanyway, let's not get carried away17:37
mgedminI'd be happy to participate in a discussion about modern revision control systems17:38
mgedminbut after the meeting17:38
mgedminum... are we still in the meeting?17:38
th1aI'll send out an email on these issues.17:38
* th1a bangs the gavel.17:38
mgedminI'd buy a beer/$other_beverage if someone wrote a Supybot plugin for announcing meeting start/end times on IRC17:39
th1amgedmin:  They don't really change.17:40
mgedminfor some reason I can never remember the new time17:40
srichter13:30 UTC! :-)17:40
srichterand it's in the title of the channel too :-)17:40
srichterokay, anyways, I have to go back to my decision making about my TA17:41
* mgedmin enters the time into his PDA17:41
* mgedmin wonders why he hadn't done that before17:42
th1amgedmin:  I have started using your time tracker from the command line on my Mac.17:42
mgedminyou mean
th1aI don't know whether I should write a gui for the Mac or my phone.17:43
th1amgedmin: yeah.17:43
mgedminMac GUI would be awesome17:43
mgedminAiste would be happy17:43
th1aI'll keep that in mind.17:43
mgedminis there a reason why you do not use the Gtk+ one with Apple's X server?17:43
th1aWell, I took a poke at it.17:44
th1aProbably it would work if I bothered to figure out how to point pygtk to the DarwinPorts GTK.17:45
Aistehm, it works for me17:45
srichterBTW, someone reported a failure on the refactor branch17:45
Aistethere is a py-gtk2 for darwin ports17:45
srichterI just ran all the tests again and they pass for me17:45
th1aAre you also using DarwinPort's Python?17:46
ignassrichter, pdf generation is failing :/17:46
ignashow many unit tests does test runner display ? (the count)17:47
srichterbut I have the directories all setup and it does not tell me that the tests are skipped anymore17:47
th1aI had been trying to keep Apple's Python as the main one on my machine to make sure that SchoolTool worked properly with it.  But now that it seems like those problems are gone, I guess it doesn't matter.17:47
ignasok, i'll run them once more17:47
srichterwhich test in particular is failing?17:47
Aisteth1a: yes, I am using python23 from DP17:48
ignassrichter, let me see, it was a long time17:48
srichterignas: oh yes, I fixed the PDF tests last, because I never had the right directory setup17:49
ignasImportError: No module named http17:49
ignasthat's what i am geting :/17:49
ignas  File "/home/ignas/src/schooltool-r/src/schooltool/app/", line 48, in ?17:50
th1aAiste:  After I put together a rough .pkg installer for Mac last week, I got a bunch of help from an Apple Education trainer, and he vastly improved it, so it has a controller in the preference pane for the server & everything.18:03
srichterthen you are in troubles :-)18:04
srichterignas: because this is a Zope 3 import18:04
srichteris Zope 3 checked out for you?18:04
srichterdid you get a fresh checkout?18:04
Aisteth1a: cool18:05
th1aI'll let you know when you can test it.18:06
*** muzz2k has joined #schooltool18:06
*** muzz2k has left #schooltool18:06
mgedminyipeee, I finally got the infamous nondeterministic ConnectionStateError myself!19:26
mgedminand it is deterministic19:29
th1aWhat's the cause?19:31
mgedminpython -f -w1 -pvvC -s src/schoolbell/app/browser . 'copypas|errors'19:31
ignassrichter, i have just rm -rf ed my Zope319:31
ignasand still - i am geting the import error19:31
* mgedmin updates
mgedminsrichter, watch this:19:38
mgedmincd ~/src/srichter-refactor/schoolbell19:38
mgedminsvn up19:38
mgedminsvn up Zope 319:38
mgedminmake test ftest19:38
mgedminfails with19:38
mgedmin    ConfigurationError: ('No such file', '/home/mg/src/srichter-refactor/schoolbell/Zope3/src/zope/app/generations/browser/Zope3/src/zope/app/generations/browser/')19:38
mgedminwhile doing "Extracting translations"19:39
mgedminsrichter, LC_ALL="C" python -w1 -pvC gives me 11 failures19:40
mgedminall in
mgedminsvn st shows no local changes19:41
mgedminI will set up buildbots for your branch19:41
srichterthe above directory does not even exist! :-)19:41
srichterbtw, forget the schoolbell directory on the branch19:42
srichterit should simply be removed19:42
ignassrichter, and what about schooltool ?19:42
srichtereverything should be passing in it19:42
ignascould you please do "svn co svn+ssh://" + make test ftest19:43
mgedminah, ok, forgetting19:45
mgedmin$ svn rm schoolbell19:48
mgedminsvn: Use --force to override this restriction19:48
mgedminsvn: 'schoolbell/Zope3' is not under version control19:48
srichteryou have to remove Zope3 normally19:48
srichterbecause it si a checkout19:48
srichternot part of the revision control19:49
srichterbut it has svn entries, because it is a checkout19:49
mgedminoh, right -- but before svn told me that, it went on to list 49 "Performing status on external item at 'schoolbell/Zope3/src/foo'" lines19:49
mgedminit confused me19:49
povbot/svn/commits: * mg committed revision 4948:19:52
povbot/svn/commits: Removing decoy.19:52
povbot/svn/commits: <srichter> btw, forget the schoolbell directory on the branch <srichter> it should simply be removed19:52
ignassrichter, so how as the clean checkout ?19:52
srichterok, now I get the http problem as well19:53
ignasnow it's your problem :D19:53
srichteroh, and I know why :-)19:53
srichterit's from the changes Jim and I did last week I think19:54
mgedmin$ make ftest gives me FAILED (failures=5)19:54
mgedminunit tests all pass here19:54
mgedminyay, I'm glad we all reached the same state19:54
mgedminpostponing buildbot change until these bugs are resolved19:55
srichterso we might have to migrate to WSGI19:57
srichterprobably trivial19:57
srichterI can look at it tonight19:57
srichterright now I am really busy19:57
mgedminwe probably won't be able to start on the merge today19:59
th1aAiste:  Do you have gtimelog running under GTK on Mac?20:22
mgedminshe's not here20:23
povbot/svn/commits: * mg committed revision 4949:20:23
povbot/svn/commits: Removing superfluous ftest -- this should fix issue377 (ConnectionStateError).20:23
th1aDo you know if she does?20:23
mgedminyes, she does20:23
*** alga has joined #SchoolTool20:24
povbot/svn/commits: * mg committed revision 4950:20:29
povbot/svn/commits: Oops.  Apparently now I have to edit some ZCML when I remove a ftest.20:29
povbot/svn/commits: * mg committed revision 4951:20:30
povbot/svn/commits: Unused imports.20:30
povbot/svn/commits: * mg committed revision 4952:20:33
povbot/svn/commits: Don't Repeat Yourself: refactored the hasPermission implementation.20:33
povbot/svn/commits: * ignas committed revision 4953:20:35
povbot/svn/commits: Refactored common code into ACLViewBase.applyChanges20:35
povbot/svn/commits: * mg committed revision 4954:20:37
povbot/svn/commits: Removed dead code.20:37
mgedminhasPermissions([perm1, perm2, perm3]) -> [True, False, True]20:40
mgedminhasPermissions([perm1, perm2, perm3]) -> [perm1, perm3]20:40
mgedminanswer: #1 (because we're doing optimization, and do not want to do __contains__ tests in a loop)20:41
* mgedmin can't even spell duh!20:41
*** tiredbones has quit IRC20:49
povbot/svn/commits: * ignas committed revision 4955:20:58
povbot/svn/commits: Added a unit test for ACLViewBase.permsForPrincipal20:58
povbot/svn/commits: * mg committed revision 4956:21:01
povbot/svn/commits: Refactor ACLViewBase to use hasPermissions instead of hasPermission.  Ignas claims this gives a massive speedup.21:01
povbot/svn/commits: * mg committed revision 4957:21:04
povbot/svn/commits: Removed hasPermission which is now no longer used.21:04
th1aHm.  While trying to get gtimelog running I discovered a DarwinPorts bug report by Aiste.21:19
Aisteyeah, I do not know whether that bug is still valid -- do not have time to install python24 from scratch and then all the py-gtk stuff and so on21:21
Aisteit did not work the three times I tried it21:21
algaserves you right for choosing such a user-oriented platform :-)21:25
*** thisfred has quit IRC21:29
Aistealga: huh, what does DarwinPorts have to do with user oriented?21:30
algaMac is for grannies :-)21:31
povbot/svn/commits: * mg committed revision 4958:21:33
povbot/svn/commits: Removed unused import.21:33
Aistealga: hello to you too21:33
*** srichter has quit IRC21:36
*** ignas has quit IRC21:51
th1aAiste:  So I still have to manually compile pygtk?21:53
Aistewell I did not21:54
povbot/svn/commits: * mg committed revision 4959:21:54
povbot/svn/commits: Added a script to measure ACL view rendering and update times.21:54
Aisteand I use python2321:54
Aistemy py-gtk gets installed against python2321:54
th1aThere doesn't seem to be a py-gtk port.21:54
Aistewhich is ok with some exceptions21:54
th1aOnly py-gtk2.21:54
Aisteyes, that is the one that I am using21:55
Aistewith glade variant21:55
mgedminhow do you write native Mac OS X apps with Python?21:56
th1aIn this case I think PyObjC would be the best choice.21:57
th1aOr wxPython!21:57
mgedminACL view rendering: 5 seconds (rev 4937) -> 2.9 seconds (rev 4959)21:57
mgedminACL view form processing: 10 seconds (rev 4937) -> 5 seconds (rev 4959)21:57
mgedminthat's with 100 random users and groups21:57
mgedminon my Pentium-M 1.6 GHz laptop21:58
th1aWith batching?21:58
mgedminalthough batching is only in the UI21:58
mgedminthe view loads and computes stuff for all persons & groups in the backend now21:58
mgedminthat's the next item on the list21:58
th1aI see.21:59
th1aSo if it only calculates the objects its displaying it'll be several more times faster?21:59
mgedminI hope so22:00
mgedminlet's see...22:00
*** appyR has quit IRC22:11
mgedminspike results: with proper batching22:16
mgedminACL view rendering: 0.35 seconds22:16
mgedminACL view form processing: 0.5 seconds22:16
mgedminwill write tests and commit it tomorrow22:16
mgedminsee ya!22:16
*** mgedmin has quit IRC22:16
*** alga has quit IRC22:53
*** srichter has joined #schooltool23:09
*** maharaja has joined #schooltool23:33

Generated by 2.15.1 by Marius Gedminas - find it at!