IRC log of #schooltool for Tuesday, 2006-06-06

*** Aiste has quit IRC00:04
*** Aiste has joined #schooltool01:03
*** jinty has quit IRC04:13
*** strichter has joined #schooltool05:14
*** srichter has quit IRC05:14
*** _srichter has joined #schooltool05:18
*** _srichter is now known as srichter05:18
*** strichter has quit IRC05:39
*** strichter has joined #schooltool05:56
*** srichter has quit IRC05:56
*** strichter is now known as srichter05:56
*** th1a has quit IRC06:08
*** th1a has joined #schooltool06:17
*** th1a_desk has joined #schooltool09:52
*** th1a has quit IRC09:52
*** wimbou has joined #schooltool10:01
*** Aiste has quit IRC11:01
*** kitblake has joined #schooltool11:22
*** Aiste has joined #schooltool11:28
*** thisfred has joined #schooltool11:55
*** faassen has joined #schooltool12:09
*** mgedmin has joined #schooltool12:18
*** ignas has joined #schooltool12:22
ignasfaassen: hi12:37
povbot/svn/commits: * faassen committed revision 6157:13:13
povbot/svn/commits: Move index.html and edit.html from schooltool.person under the control of schooltool.demographics. These now simply redirect to nameinfo/index.html and nameinfo/edit.html, whereafter the user can tab through the data.13:13
povbot/svn/commits: This of course resulted in the editing of a quantity of tests as well.13:13
*** jinty has joined #schooltool13:17
*** alga has joined #SchoolTool13:18
ignasfaassen: are you there ?13:21
povbot/svn/commits: * faassen committed revision 6158:13:50
povbot/svn/commits: Make sure we grab zc.resourcelibrary 0.5.1. This has the deprecation warning fixed.13:50
faassenmassive urgency concerning the deprecation warning, huh?14:27
povbot/svn/commits: * faassen committed revision 6159:15:24
povbot/svn/commits: Add a bit of explanation about this function.15:24
povbot/svn/commits: * faassen committed revision 6160:15:26
povbot/svn/commits: Two blank lines between top-level declarations.15:26
*** alga has quit IRC15:28
*** alga has joined #SchoolTool15:32
*** pcardune has joined #schooltool15:58
povbot/svn/commits: * faassen committed revision 6161:16:00
povbot/svn/commits: Add a few simple docstrings.16:00
*** gintas has joined #schooltool16:12
*** alga has quit IRC16:18
*** alga has joined #SchoolTool16:28
*** th1a_desk is now known as th1a16:30
* th1a shuffles some papers around.16:31
th1aHi everyone.16:31
th1aTime to start the meeting.16:31
* mgedmin shuffles in his seat16:32
mgedminhi all16:32
th1afaassen, srichter?16:33
* gintas is here too16:33
th1ahi gintas.16:33
faassenI'm here.16:35
th1aAh, good.16:35
th1aI had a long chat with srichter last night.16:35
th1aI forgot to ask/remind him about the meeting though.16:35
th1aBut he's stuck working on some other stuff until Friday anyhow.16:36
th1aSo... let's start with POV this week.16:36
th1aHow are things coming?16:36
* mgedmin shifts the spotlight to ignas16:38
ignaswe are still filling up the table with white X marks16:38
ignasand replacing commented out tests with new ones16:38
ignasi think either tomorow or on thrusday we should be ready for a merge16:38
th1aIt seems like things are coming along smoothly.16:39
ignasthough security will not be completed it will be usable enough not to hinder the progress of other developers16:39
ignaswhich was the reason of developing it in a branch16:39
th1aHow much will the rest of the code need to be changed?16:40
ignaswhy should it be changed?16:40
ignasunless there were some deprecated use of permissions that will cease to exist soon in the zcml16:41
gintaswe'll probably just touch ZCML in a few places16:41
ignasthough we are hunting these down ourselves16:41
faassenignas: (earlier non-security policy approaches would've caused rather large changes in the codebase, so this is probably why th1a asked :)16:41
gintasand perhaps add a few extra crowds if needed, but that's a small problem16:41
gintasfaassen: indeed16:41
*** erchache has joined #schooltool16:42
gintasthis job of porting security is a bit tedious, but I'm so happy we didn't continue the down subscriber route16:42
th1aI'm just checking to make sure there aren't unpleasant surprises coming my way.16:42
gintass/the down/down the/16:42
erchachehi everybody :-D16:42
erchacheanybody can say me state of development of project please?16:43
ignaserchache: the state is "YELLOW"16:43
erchachei dont touch schooltool since last update :-(16:43
algaerchache: we're in the meeting right now16:43
erchachesorry :-S16:43
th1aYes, I'll update you when we're done erchache.16:43
ignasi am considering most changes to the codebase as a part of the access control story16:44
erchacheok...i can wait....sorry16:44
erchachei dont remember today is tuesday ;-P16:44
th1aOK, so remind me what POV's planned deadline is for this contract.16:44
ignasi wanted to ask you the same question16:45
th1aI guess one of us has to look it up then ;-)16:45
*** srichter has quit IRC16:45
ignaswe haven't decided upon it i think, you told me that it *MUST* be finished before 06-1716:45
th1aWell, is that what you've been working towards.16:47
ignasso i was aiming for at least 2-3 days before that to have some time for emergencies (those will probably not happen ...)16:47
th1aWell, let's leave 6/16 as the deadline, then.16:48
th1aAny other questions, problems, etc., from the POV side?16:48
faassenjust me. :)16:49
faassenI'm a question, a problem and an etc for the POV side sometimes. :)16:49
ignasdo you like gnumeric or open-office spreadsheet ?16:49
th1aI guess I prefer Open Office.16:49
th1aI don't like spreadsheets in general.16:49
ignasth1a: well they are way better for tables than open office writer16:50
th1aOh, yes, I suppose that's true.16:50
ignasth1a: if you would get a couple of clayvorance devices it would make everything so much easier for inter POV-faassen communication ;)16:50
faassenyes that would be useful.16:51
th1aignas:  There are moments when my background as an English major comes out, like when I do tables in a word processor instead of a spreadsheet.16:52
th1aOK.  Go ahead, faassen.16:52
ignasth1a: hey, but you can spell clairvoyance properly :)16:53
th1aignas: I checked to make sure.16:54
gintasignas, you made my day ;)16:54
ignasgintas: yes, yes i know it's not exactly accurate16:54
faassenclayvoyage is at the level of my gravel thing. :)16:54
faassenclayvorange, that is.16:55
faassenclairvoyance would be easy.16:55
faassenanyway, pupeno and I spent most of last week moving bits of view logic around.16:55
faassenand fixing tests when we broke all of them again.16:55
faassenI went by half the functional tests in schooltool to make addPerson into a functional.16:56
faasseninto a function, that is.16:56
faassenso we could easily add a required field to it.16:56
faassenwe also made a few more drop-down lists work properly in the demographics screen.16:56
faassenhopefully today I can finally finally move the table.html over to index.html16:57
faassenfix a lot more tests.16:57
faassenand then we're sort of done, pending feedback.16:57
faassenI can imagine all kinds of UI feedback.16:57
faassenit's just I suppress the urge to actually do much about it.16:57
faassenuntil we get actual UI feedback.16:57
th1aI did send you an email on Sunday.16:57
faassenhm, I think I missed that.16:57
faassenyesterday was a holiday and I'm not all caught up yet.16:57
faassenokay, sorry.16:58
faassengot it.16:58
faassenwe'll proces that.16:58
th1aMy only significant concern is that we've sort of left out the "add" process.16:58
faassenwe can put more things in the add form.16:59
faassenbut we don't have a multi-screen add workflow.16:59
faassenmaking that work would impact a lot of tests.16:59
th1aWe'll have to think about how we want to do that.16:59
faassenthough I guess my refactoring would fix most of that.16:59
th1aBasically the "name" screen should be enough to instantiate a new student.17:00
faassenanyway, for an add workflow I'd really like to see some UI design somewhere.17:00
faassenwe could reuse more of the name screen.17:01
th1aOK.  I can send you something.17:01
faassennote that the tests make a lot of assumptions of where you end up after adding a peron.17:01
faassenI mean, tests in schooltool in general.17:01
th1aOy vey.17:01
faassenI guess I can fix that now in a central place though.17:01
faassenby simply making sure that after the add procedure it ends up in the place the rest of the ST tests expect.17:01
faassenanyway, thanks.17:01
faassenit's possible that today's checkin by me already changes some of the workflow.17:02
faassenyou may want to check that out.17:02
th1aOK.  I'll have a look at it.17:02
faassenas the behavior of the edit and index presentation changed.17:02
faassenanyway, I also just had a good discussion with ignas17:03
th1aI'm really feeling like test brittleness is a big problem in our codebase.17:03
faassenhe patiently waited until I stopped being overly defensive and I am adding some documentation to the ocde.17:03
th1aGood man, ignas.  Patience is a virtue.17:03
faassenhe was missing some tests in the demographics package, though I think I made at least somewhat clear that the demographics package doesn't really add much in the way of APIs or algorithms.17:04
faassenso there is relatively little to do in the way of unit testing, and more integration testing (which is best done by functional tests)17:04
faassenI think that's also part of my frustration with the package. it's not doing anything really, just storing stupid data, but the integration work costs a lot of time nonetheless.17:04
faassenand I don't feel like I'm designing a cool new APi or something.17:05
faassenanyway, I'll try to add a few tests in some places he pointed out.17:05
faassenand a document sketching out the design of the demographics package.17:05
faassenso that people have at least a bit of an idea of how it is put together.17:05
th1aOK.  That would be great, faassen.17:05
faassenI must admit I exploded, I just spent forever fixing existing schooltool tests and then I heard from ignas I didn't do enough with tests. :)17:06
* faassen grins.17:06
faassenbut the clairvoyance we got working helped.17:06
faassenanyway, I will focus today on getting the table.html into index.html move done.17:07
faassenrest of today. probably bit of tomorrow.17:07
faassenand handling Tom's UI feedback.17:07
faassenplus the documentation bits.17:07
th1aI wish I knew what we should do to make it easier to make small changes in code or presentation without blowing up lots of tests.17:07
th1aOK, thanks faassen.17:07
faassenand then hopefully by thursday or friday we can conclude the demographics for the time being.17:07
faassenwell, I whined a bit at Jim the other day, and he said knowing what to test is an art.17:08
faassenfunctional tests are difficult that way.17:08
faassenespecially browser tests, when you go mess about with the UI as we did.17:08
faassenit wasn't helped that the person package is used everywhere as the sort of, example package.17:08
faassenso devmode broke.17:08
faassenas it was using the person package.17:08
faassenin some tests.17:09
faassenplus in other testing code it's used to set up users for security tests.17:09
faassenso again, boom.17:09
faassenthe big refactoring of all those tests to make the person setup more central should help at least somewhat.17:09
faassenth1a: oh, one larger open issue for demographics is evolution.17:10
faassenI still hope I can conclude the demographics work by friday. I'll try to get an evolution script going.17:10
faassenthat takes a february schooltool version17:10
faassenand gets it to a current checkout, orsomething like that.17:10
th1aThat would be the goal.17:10
faassenthough that'll require some fiddling.17:10
th1aI think faassen will deserve a SchoolTool merit badge at the end of this.17:11
th1a"I survived my first SchoolTool contract."17:11
* faassen grins.17:12
faassenwell, it's all downhill from now. :)17:12
faassenexcept for all those trees.17:12
faassenand the patches of ice.17:12
faassenand so on.17:12
faassenbut all downhill.17:12
th1aWe'll definitely have to debrief and some point and talk about what will make it easier for the next person.17:12
alga"Schooltool is going downhill" -- faassen   :-)17:12
* faassen laughs.17:12
faassen"Since I got involved with it, Schooltool is going downhill rapidly!"17:13
th1aSo it looks like various delays have at least put things back in sync, insofar as all the current contracts will be done at the same time more or less.17:13
th1aSo I'm going to announce a release I guess on the 20th, rather than, um, last week.17:14
th1aI'll have to think about what to call it...17:14
th1aI guess it should still be alpha2.17:15
th1aMy goal for today is to set up a good attendance demo.  Attendance has been difficult to demo.17:15
th1aI finally decided that I need to create a school with 10 students and one teacher.17:16
th1aThat meets every day of the year.17:16
th1aFor most of or all of the day.17:16
th1aI think that will be make it easier to test.17:16
algaso that 1 person per period can come and try it out?17:16
th1aWell, right now it is hard for one person to get a sense of what's going on in the 1000 person demo school.17:17
th1aI mean, if I mark a student absent one period, and I want to see if that shows up the next period, I have to figure out what other classes they have that day, etc.17:18
algaaaaa, that's realistic sample data for you! :)17:19
th1aI need to generate some useful attendance feedback.17:19
th1aI'm certainly glad we're doing the big sample data now.17:19
th1aOK.  Anything else?17:20
th1aAll right.  Have a great week, folks.17:21
th1aYes gintas?17:22
gintaswhat's the plan on increasing schooltool's performance?17:22
gintasI don't think it is adequate at this time17:22
gintasat least I feel the pauses pretty badly on my computer17:22
erchacheth1a: can upgrade from schooltool 0.11.4 to schooltool 2006 svn?17:23
gintasadmittedly it's a bit outdated (that's why the pauses aren't so obvious to the others)17:23
erchachei want to get the lastest stable version17:23
th1aerchache:  Well, svn is sort of by definition not stable.17:23
erchachecan or not?17:24
gintasbut in deployment the server will have many simultaneous users17:24
th1afaassen just said he was going to try to get the evolution scripts written at the end of the week.17:24
gintaserchache: the upgrade is not guaranteed to go smoothly17:24
th1aSo the answer is definitely no this week.17:24
gintasyou may lose your data17:24 long is spected to convert into stable?17:24
gintaserchache: I think Tom planned a release on the 20th, right?17:25
th1aYes.  Try it then and let us know what breaks.17:25
erchacheyepa....its ok for me17:25
gintasso, th1a, any thoughts on that?17:25
gintasI am a bit afraid of this issue because as far as I know no one has done any quantitative measurements17:26
th1aWell, I don't know.  When non-Lithuanians bring up performance they tend to get jumped on.17:26
pcarduneI think the biggest factor slowing schooltool down is the relationship package.  (at least that is what I seem to remember)17:26
gintaspcardune: we have optimized it a bit recently17:26
gintasit should not be that big of a deal17:26
th1aWe just have to see, I think.17:27
th1aIt runs like crazy on my new laptop.17:27
pcarduneth1a: is that a good thing or a bad thing?17:27
gintaswell, think how crazy it would run for 100 pupils checking their timetables in the morning17:27
* jinty throws an AJAX grenade in the conversation and runs;)17:28
gintasMaybe I'm just making unnecessary waves, but my point is that we should do some measurements.17:28
th1aI would be in favor of that.17:28
gintasBecause if we have a serious problem here, we will not be able to solve that overnight.17:29
pcarduneI think the relationship package should be refactored anyways.  I don't understand it half the time.  I'd like to see some relationship zcml tags.  That would be cool.  But those are just my 1.5 cents17:29
th1apcardune:  Your input is very welcome.17:29
gintaspcardune: that sounds interesting17:29
th1aStrongly encouraged.17:29
gintasnice idea17:29
th1agintas:  I think it would be a good thing if POV came up with some performance metrics.17:30
ignasth1a: are tou sure ? :)17:31
th1aOne of my few frustrations with the POV approach is that you are so dogmatic about not optimizing prematurely, it means that optimization becomes a separate (paid) task.17:32
ignasthough yes indeed we should resurrect our benchmarking facilities17:32
th1aWhen perhaps it should have been part of the original work.17:32
gintasth1a: it is a bad idea to compromise clarity for performance early on17:33
ignasand it costs more in the long term17:33
gintasit is also a bad idea to optimize without measurements17:33
th1aI think we can all agree that measurements would be good.17:33
th1aOK.  Our time is up here.17:34
* th1a bangs the virtual bag of gravel.17:34
th1aHave a good one folks!17:34
th1apcardune:  It is important that you let us know what you're thinking.17:34
pcardunewell, I was just thinking in passing17:35
gintasth1a: generally optimization works out pretty well as a separate task17:35
th1aYou're not an outsider in this process.17:35
pcarduneI don't really use the relationship package much myself, and I wouldn't put it as the highest priority17:35
gintasI guess that's one of the reasons we did not do it much17:35
th1agintas:  As another developer, I agree.  As your client, I don't.17:36
gintasit did not directly contribute to goals defined in the stories17:36
faassenconcerning performance, just do an eyeball test..17:36
gintasI see your point17:36
faassenby clicking through the edit screens.17:36
faassenand when you hit schooldata17:37
th1apcardune:  Well, yes, ultimately we end up at needing benchmarks and profiling to know what needs to be done.17:37
faassenwe added some code there.17:37
faassenand that screen is veyr much slower.17:37
faassenand I believe that's because it pulls in some relationship stuff there.17:37
gintaseyeball tests are OK for order-of magnitude changes17:37
faassenwhen large amounts of users are there.17:37
faassenyeah, eyeball tests are a good start to identify "this screen is slow"17:37
gintashowever, I think it is a good idea to pinpoint the exact location which wastes time17:37
gintasthat's where a profiler comes in handy17:38
th1aThere is also the question of which are the most efficient changes in terms of developer time.17:38
ignastwo parts: bechmarking a.k.a. find out which views are too slow with many users/data (can be done by anyone) needed to identify which parts need improvement17:38
ignasand profiling: done by the person who is optimizing something that is too slow, used to pinpoint the places in code that need serious optimization17:39
faassenth1a: oh, in your feedback, could you also let us know where you want to see the photo?17:39
faassenth1a: (we still haven't really cleared why it only works if you use the .jpeg extension)17:39
faassenth1a: (in the url)17:40
ignaswe need some infrastructure to make the benchmarking more scientific (add support for measuring whether something got faster, and how much faster it is)17:40
faassenI wouldn't mind a little benchmarking project. :)17:40
th1aI would guess on the page with the names.17:40
gintasin formal software engineering there are requirements such as "this view must take no longer than 0,5s to render"17:40
faassenth1a: send it in the same mail as the one that'll contain your add form suggestion :)17:40
* jinty after a bad experience, always tries to include "enough" optimization in his quotes. gintas is lucky with th1a as a client.17:40
gintasin our case it's probably too formal and strict17:40
* th1a is too soft as a client.17:41
* faassen meant to say, I wouldn't mind doing a bit of optimization. :)17:41
faassenit was at least interesting to see how much that one bit of code added to the schooldata form makes it so much slower17:41
faassen(if lots of testdata exists)17:41
faassenotherwise those forms are pretty snappy.17:41
gintasmy proposed course of action would be to find a relatively slow (sub-1 GHz) computer and do a wall-clock test to find the worst bottlenecks17:42
gintaswith schooltool running on that computers17:42
th1aRegardless, I think I'd rather do optimization later in the summer.17:42
gintasthen they can be prioritized and included as a story in a contract17:42
faassenth1a: sounds like a fun project, optimization. :)17:43
gintasth1a: I would suggest to at least do the tests early on17:43
th1aOK.  I'll keep it in mind.17:43
gintasin the past we may have been able to pick some low-hanging fruits and get 10x boosts, but there is no guarantee that we can keep it up17:43
faassenwe may go the normal route where endusers, very patient actually, will complain about stuff that's just too insane.17:43
faassenwirdly enough sometimes they do complain but you never actually hear their complaints.17:44
gintasfaassen: I think that would be too late17:44
gintasif SchoolTool is deployed in schools and it turns out that it is too slow, we're in very big trouble, I think17:44
faassenI'm just describing what normally happens.17:44
faassennormally I just try to write reasonably performant code. :)17:44
gintasfaassen: me too17:45
faassenanyway, yeah, but features not doing what the school wants..17:45
th1aOur partners won't be doing big deployments initially, where you'd have a thousand kids checking their calendars.17:45
faassenis arguably worse. it's hard to judge right now.17:45
jintygintas: but only if you deploy schooltool too fast...17:45
th1aFunctionality will be much more important for the partner schools.17:45
gintasth1a: how large are the initial deployments going to be?17:46
gintasI previously thought that we were doing a full-scale installation on 3 or 4 partner schools17:46
gintasthis autumn17:47
th1aWell... the thing is that I don't think students will be extensively using SchoolTool at any of them.  Maybe in Philly.17:47
th1aBut the admin apps are only used by teachers.17:47
gintasah, in that case indeed it's not critical17:47
th1aSo a few dozen teachers doing attendance at the beginning of the period, and teachers doing grades at the end of the semester.17:47
gintasyes, that's a piece of cake17:48
th1aI mean, I would hope a few schools somewhere will use our calendars with all their laptop carrying students, but it isn't at the top of our list of concerns.17:48
gintasI was thinking about hordes of students violently attacking our lone SchoolTool server ;)17:48
th1aSomeday ;-)17:48
gintasWell with this new information I guess I can take back my alarms.17:50
gintasthanks for the clarification17:50
th1aOK.  It is a little confusing, because calendaring is so different than the admin functions.17:51
*** whaddon has joined #schooltool17:55
*** erchache has left #schooltool18:09
*** wimbou has quit IRC18:13
ignasth1a: what do you think of having this on #schooltool channel?18:24
mgedminignas: could you elaborate?18:29
ignasby it i mean the "lisppaste" bot18:30
ignasas it has to be in the channel so it could display URLs to pasted code snippets18:30
povbot/svn/commits: * faassen committed revision 6162:18:34
povbot/svn/commits: Some docstrings, style adjustments.18:34
*** whaddon has quit IRC18:34
*** whaddon has joined #schooltool18:34
*** whaddon has quit IRC18:36
*** whaddon has joined #schooltool18:36
*** whaddon has quit IRC18:40
*** ignas has quit IRC19:09
*** srichter has joined #schooltool19:18
*** strichter has joined #schooltool19:24
*** srichter has quit IRC19:27
*** _srichter has joined #schooltool19:31
*** _srichter is now known as srichter19:35
*** strichter has quit IRC19:47
povbot/svn/commits: * faassen committed revision 6163:19:49
povbot/svn/commits: Add some documentation about design.19:49
*** erchache has joined #schooltool19:52
*** strichter has joined #schooltool19:58
*** srichter has quit IRC19:58
*** strichter is now known as srichter19:59
*** pcardune is now known as pcarduner_lunch20:00
povbot/svn/commits: * faassen committed revision 6164:20:16
povbot/svn/commits: A title is needed.20:16
*** pcarduner_lunch has quit IRC20:23
*** thisfred has left #schooltool20:23
*** pcarduner_lunch has joined #schooltool20:33
*** faassen has quit IRC20:48
*** erchache has quit IRC20:49
th1ajinty:  ayt?21:24
jintyth1a: yep22:03
*** alga has quit IRC22:13
*** srichter has quit IRC22:25
*** gintas has quit IRC22:38
th1ajinty:  I got the new server.22:42
th1aWhen do you have some free time ;-)22:42
th1apcarduner_lunch:  Some competition:
*** pcarduner_lunch is now known as pcardune22:45
*** gintas has joined #schooltool22:48
pcarduneth1a: I don't feel threatened22:51
pcardunehe he he22:51
*** srichter has joined #schooltool22:54
*** mgedmin has quit IRC23:04
jintyth1a: very cool! Er, free time;) not now, I'm cooking supper.23:04
jintylet's talk thursday, I'll probably be able to do something about it next week23:05
th1ajinty:  Sounds fine.23:06
th1apcardune:  What they're doing sounds interesting, it is closer to what I was doing at Feinstein than CanDo.23:06
*** gintas has quit IRC23:09
*** pcardune has left #schooltool23:55

Generated by 2.15.1 by Marius Gedminas - find it at!