IRC log of #schooltool for Tuesday, 2006-09-05

*** wrobel has quit IRC00:04
*** srichter has joined #schooltool01:07
*** didymo has joined #schooltool01:29
*** ignas has quit IRC02:38
*** eldafar has quit IRC02:57
*** eldafar has joined #schooltool03:00
*** jinty has quit IRC03:24
*** eldafar has quit IRC05:37
*** wrobel has joined #schooltool08:59
*** Aiste has quit IRC12:10
*** ignas has joined #schooltool12:13
*** gintas has joined #schooltool12:33
*** Aiste has joined #schooltool12:34
*** alga has joined #SchoolTool12:41
*** jinty has joined #schooltool12:41
*** Aiste has quit IRC13:42
*** vidasp has joined #schooltool14:17
*** didymo has quit IRC14:29
*** faassen has joined #schooltool14:39
*** mgedmin has joined #schooltool15:18
*** gintas has quit IRC15:50
*** Aiste has joined #schooltool15:53
*** Aiste has quit IRC15:55
* th1a shuffles some papers around.16:31
th1ahoi ignas?16:32
ignasvidasp: are you in there ?16:33
th1aSo, what's going on in Vilnius?16:34
ignasi should finish the proposal either today evening or tomorrow morning16:35
ignasstumbled on a couple of Zope3 bugs last couple of days16:35
ignasbut now they seem fixed so buildbod should stop shouting and i will commit some more of the calendaring with new UI stuff16:36
ignasvidas didn't do that much since we last talked as he was busy on friday16:36
ignasand he is not working on Mondays and Tuesdays16:36
th1aAre you working on the trunk now?16:37
ignasno, on a branch16:37
th1aOK, I'll check that out.16:37
ignasso i would not disrupt jinty's or anyone elses work16:37
ignasas soon as Vidas will do the calendar widget calendar views will be finished up and i'll work on attendance16:38
ignasi have Proposal, Calendaring, Attendance, Temporary tabs in the order of priority16:39
th1aAh, right.16:40
th1aI've been trying to figure out how we're going to handle this transition to your working full time.16:40
ignasas for schools, as soon as gintas comes back (he's away for the rest of the week) we'll try to contact a school16:41
th1aOK... Do you have any in particular in mind?16:41
ignasjust to find out whether that particular school (the one gintas studied in) is interested16:41
th1aIs that a high school?16:41
ignasVilnius Lyceum16:42
ignashigh school16:43
th1aDo they do teacher education at the University?16:43
ignasteacher education ?16:44
th1atraining people to be teachers.16:44
th1aIf so, that might be a good place to get leads.16:45
ignasnot sure about it16:47
ignasif we didn't need one on one contact16:47
ignasi could get a school easily as my mother is informatics teacher, but the schoolt is 350 km away from Vilnius16:48
th1aHm... well, still, that is good to know.16:49
th1aThat's still in Lithuania?16:49
ignasyes :)16:50
ignasmy hometown16:50
th1aIf we couldn't find anywhere else, would it be a possibility for you to work with them and visit a few days a month?16:51
th1aI don't know how hard it is to get there, etc...16:52
ignasmy mother would be happy about that :) but i'd rather stay here (the ticket both ways is 10 Euro for a student and 20 Euro for a grownup)16:52
ignasbut still vlinius would be much better ;)16:53
th1aI'm sure you'd rather stay in Vilnius.  I've just been trying to figure out whether to make a proposal to Mark before you have a school.16:54
th1aI think we'd both just be afraid that it might never happen.16:55
th1aOn the other hand, I'd like to have things settled.16:55
th1aSo are you planning on continuing to work on SchoolTool for the next few weeks, or are you going to have other work to do?16:57
ignaswas gone a bit talking to gintas about contact school17:02
ignaswe'll try to talk to the school before next tuesday IRC meeting17:03
ignasjust to konow whether they are  interested17:03
th1aOK.  That would be helpful.17:04
ignasi am pretty sure that i'll be working on schooltool for next 2-3 weeks17:04
ignasoh, as for the full time concept, I hope that you will not have a problem with a plan to "borrow" some time from POV to do pair programming on more difficult parts of design/implementation17:06
th1aNo, that is a good idea.17:06
ignaswhile returning those hours by working on POV projects for that much time17:06
ignasto even it out17:06
th1aThe important part is giving you more of a sense of ownership and responsibility for the project as a whole.17:06
th1aAnd not routing all the communication through me.17:07
th1aAnd not having me make technical decisions...17:08
*** gintas has joined #schooltool17:08
ignasthat would leave a lot less thinking about "should i be doing this, and who is paying" for me i hope ...17:08
th1aYou'll have more responsibility for deciding what needs to be done to get ST working at your school.17:09
th1aI'll have a few priorities, too.17:10
th1aBut I'll try to limit that.17:11
th1aAnd also, I'm hoping not working on a contract basis will make it easier to handle bugfixes quickly.17:11
ignaswell there will be some balancing between your priorities, project priorities (benchmarking framework, refactoring etc.) and client priorities (need that button in there) ...17:12
ignasbut i think it is manageable17:12
* ignas is looking forward into the bright future17:13
th1aYes.  We may need to defer framework priorities to focus on client priorities.17:13
th1aI think we've historically spent too much time fussing with the framework.17:13
th1aIt is hard to resist.17:13
th1aWithout an anxious client.17:14
ignasno, not really, ivija is way less frameworky as it has clients using it17:14
ignasour inside project17:15
th1aWe do have an excuse with SchoolTool as our real client (Mark) explicitly wants a framework.17:16
*** th1a has quit IRC17:16
*** th1a has joined #schooltool17:16
* faassen listens in to the dicussion. :)17:17
ignasjust IMVHO but how many frameworks built from scratch (without building a few working applications or broken frameworks first) do you know?17:17
th1aHi fassen.17:17
faassenanyway, I don't see how ignas and th1a there said 'no not really' :)17:17
faassenI mean, th1a said, we may do too much frameworky stuff as we don't have clients.17:18
ignasfaassen: i was not there17:18
faassenand then ignas said, no not really because on our internal project is less frameworky as it has clients.17:18
faassenand I thought huh, they agree but why the not really? :)17:18
ignasno not really was attributed to "We may need to defer framework priorities to focus on client priorities."17:19
th1aActually, we're just a weird case because really our client is Mark, and his goal is to have a framework.17:19
ignasas i think that framework building and anxious clients are inseparable ...17:19
faassenyou need clients.17:19
faassento get a *good* framework.17:19
faassenthough of course with clients you also get a 'awwwh this framework sucks, I need to fix it'17:19
faassenso it's never perfect. :)17:19
ignas"rome was not build in one day"17:20
ignas"creating good software takse 10 years"17:20
th1afaassen: Yes, I think we've learned that clients are a necessity.  It is one of the many obvious things we've managed to demonstrate.17:20
faassenRome was not designed up front for its inhabitants. :)17:20
faassenit actually was in use straight from the start. :)17:20
th1afaassen: I was just wondering if Infrae had done that Zope 3/LDAP project yet.17:21
faassenth1a: no, still waiting for client, but Very Soon Now. :)17:21
faassenth1a: client is taking forever getting their LDAP ready17:22
th1aAlso, I'm really impressed with lxml's XML Schema validation.17:22
faassenth1a: but we don't mind as we're busy enough.17:22
faassenth1a: oh, thanks. it's actually libxml2 that does that. but lxml makes it usable for normal human beings such as you and me. :)17:22
* mgedmin is not impressed with libxml2's segfaults while processing xpath queries17:22
faassenmgedmin: it does that?17:22
mgedminask ignas17:23
mgedminhe told me it does17:23
th1aI figured the actual work was being done by libxml2, but you have made it easy.17:23
faassenmgedmin: hm, the libxml2 people are quite good at responding to bug reports.17:23
faassenmgedmin: might be a good idea to report it.17:23
ignasfaassen: ok i'll try to do that next time17:23
* mgedmin stops and wonders why he never thought about that17:23
mgedminoops ;)17:23
faassenI know you thought about that.17:24
faassenbut did you actually do it? :)17:24
* faassen grins.17:24
mgedminI saw some segfaults myself a long while ago, when I passed invalid xpath expressions17:24
mgedminit never occurred to me to send a bug report17:24
faassenokay, I don't recall seeing that with lxml, but I haven't stress tested it a lot.17:24
faassensometimes it segfaults iwth the Python bindings (not lxml) because you get a NULL back or something weird.17:24
faassenand you send it back into the code that would work with a legitimate expression.17:24
faassenand then it fails in there with a segfault due to bad input.17:25
faassen(it could happen with lxml, it's just that we consider it a bug, and the default Python bindings dont' :)17:25
th1aI have been doing a lot of work on my SIF server:  http://tinyzis.org17:25
th1aIt is almost done.17:25
th1aAdding XML schema validation of all incoming and outgoing XML has been really helpful in making things conform to the spec.17:26
th1aAnd being able to do that with a couple lines of code was great :-)17:26
faassenth1a: cool, I like small bits of code that do lots of neat stuff. :)17:27
th1aThat's what Python is all about.17:27
faassenas a tip concerning project layout, you might want to consider namespacing all your packages and put it into a 'src' directory.17:27
faassenthat makes it easier to use distutils and make an egg later.17:28
th1aYes... there is still time to rearrange that stuff.17:28
faassenlike make a package 'tinyzis' with an empty and put the source in there.17:28
th1aOne goal of this project for me is to apply YAGNI to the organization of the project.17:28
faassenit helps to separate distribution directory from package directory, basically.17:28
faassensure. :)17:28
faassenbut there's the other YAGNI: You Are Gonna Need It :)17:29
th1aThat is, I can get myself caught up at the beginning in organizing things, setting up a repository, etc., and then never write any real code.17:29
faassensure, I'm telling you now. :)17:29
ignasth1a: i did some experimentation with realtime compression of ZODB and i think that it is possible to decrease the size of schooltool's ZODB that way if we'll run into space problems with attendance tracking for a few years17:29
th1aI'll make a package prior to 0.1.17:29
faassenth1a: there's time to namespace things and organize the repos. you shouldn't start out making packages. :)17:29
th1aignas:  What kind of compression?17:29
faassenth1a: start out iwth a module, expand into a package from there, etc.17:29
th1afaassen: Right.17:29
ignasth1a: do you remember when Data.fs file was getting into gigabytes when sample data was being generated17:30
faassenI remember that taking a long time. :)17:30
th1aignas: Yes.17:30
faassenand then my harddrive being full. :)17:30
ignasth1a: so i got an idea that such things can get compressed easily, so i managed to prototype a working compressed ZODB backend (meaning that all the data is getting zlib.compressed before getting written into the hard drive)17:31
faassenignas: it's interesting that the average transaction is big enough to really benefit from compression17:31
faassenignas: like, if you change the title of some object, would that still compress? I guess it depends on the size of the object's pickle.17:32
ignasfaassen: i am not compressing transactions yet, i am compressing smalled 8kb chunks17:32
faassenignas: ah, sorry, I misunderstood. blocks of what?17:32
ignasthe prototype was dumb indexed compressed file17:33
ignasyou do zlifile.write("lot's of stuff")17:33
faassenso you store in a compressed file instead of in a non-compressed file.17:33
faassenokay, that sounds like it could work, easier and simpler, I guess. :)17:33
ignasit was a proof of concept so i did some 5-6 hours of coding17:33
faassenso how much smaller did it get?17:33
ignas~ 5 times17:34
faassenpretty good.17:34
faassenI wonder how that works for binary data like pdfs and such.17:34
faassenpeople love bloating ZODB's with those.17:34
ignasit can get better  + faster then that just that i don't have time for hobby projects :)17:35
th1aThis would, of course, be the danger of putting you on ST full time without a local client.17:35
ignasfaassen: i think you could take the diff and reimplement it yourself in a weekend to work better ;)17:36
ignasi have unit tests lying somewhere around, but "_backup" uncompressed file was more reliable on spotting bugs17:37
*** vidasp has quit IRC18:49
*** ignas has quit IRC19:35
*** gintas has quit IRC19:56
*** povbot has joined #schooltool20:10
*** faassen has quit IRC20:50
*** gintas has joined #schooltool20:59
*** mgedmin has quit IRC21:09
*** alga has quit IRC22:00
*** gintas has quit IRC23:32

Generated by 2.15.1 by Marius Gedminas - find it at!