IRC log of #schooltool for Thursday, 2005-02-10

*** alga has quit IRC00:13
*** jinty has quit IRC00:44
*** hazmat has joined #schooltool01:12
*** jinty has joined #schooltool03:03
*** jinty has quit IRC03:21
*** jinty has joined #schooltool03:21
*** jinty has quit IRC04:18
*** jinty has joined #schooltool04:24
*** jinty has quit IRC04:54
*** jinty has joined #schooltool04:55
*** jinty has quit IRC05:14
*** hazmat has quit IRC05:21
*** th1a has joined #schooltool06:00
*** hazmat has joined #schooltool06:23
*** Aiste has quit IRC06:35
*** Aiste has joined #schooltool06:42
*** hazmat has quit IRC06:58
*** hazmat has joined #schooltool07:20
*** Aiste has quit IRC07:58
*** hazmat has quit IRC07:59
*** tvon has quit IRC08:02
*** Aiste has joined #schooltool08:08
*** SteveA has joined #schooltool08:54
*** th1a has quit IRC09:07
*** Aiste has quit IRC12:28
*** Aiste has joined #schooltool12:35
*** mgedmin has joined #schooltool12:54
*** mgedmin has quit IRC14:27
*** d2m has quit IRC15:59
*** jinty has joined #schooltool16:55
*** mgedmin has joined #schooltool17:03
*** th1a has joined #schooltool18:06
th1ajinty: hi.18:06
jintyhey!18:09
jintytimetabling time?18:09
th1aSi.18:09
jintyOk, where should we start18:10
jintyHave you looked at the imaginary timetabler?18:10
th1aI've looked at it.18:10
th1aMy wife forgot to reset the alarm clock after she went to school, so I just woke up.18:11
th1aI'm a little woozy.18:11
jintyok, no worries. Then I'll just explain where I'm at18:12
th1aGreat.18:12
jintyand what I think to do next18:12
th1aSounds good.18:13
jintyOk, I tried, with the imaginary timetabler, to put something concrete on the ground.18:13
jintyA superstructure with enough inbuilt layers of indirection to cope with most problems.18:13
jintyThe fundamental concept is to store and manipulate the data on a basic level in the most generic way possible. Similar to the always store time in UTC and convert on view18:15
th1aOK.18:15
jintyminiTTML I kept separate as this is more of a general library, which I hope will be used by the solvers18:16
jintyas well18:16
th1aOK.18:16
jintyBut the major problem still is the risk. A lot is at stake if TTML does not live up to it's promise18:17
jintyalso, importing from schooltool could be difficult.18:17
th1aYes.  Actually, that brings us to something that I wanted to mention.18:17
jintygo for it18:18
th1aLooking for the URL now...18:18
*** faassen has joined #schooltool18:20
th1aGrumble grumble.18:20
jintyhe he18:20
faassenhello.18:20
th1aArticles about the thing but not the thing itself.18:20
faassencan anybody tell me how far the zope3-ification of schoobell trunk is along? i.e. enough to install it and have it do something, or not yet at that stage?18:21
th1ahttp://www.elframework.org:80/18:21
th1afaassen:  You can install it and have it do something.18:21
mgedminfaassen, you can drop the schoolbell package into PYTHONPATH, include its zcml, and then add schoolbell instances in your Zope 318:22
mgedminit is not fully functional yet, but you can play with it in Zope 318:22
jintyfaassen: don't use setup.py install - that doen't work yet18:23
th1ajinty:  As you can see from the site, the "elf" folks are trying to pull together web services and other standards related to e-learning, including a vacant slot for timetabling:  http://www.elframework.org/user_agents/timetabling18:23
* jinty goes and looks18:24
faassenI won't use setup.py install.18:24
faassenmgedmin: thanks18:24
th1ajinty:  here's a quote from an email from someone who works on elf:18:25
th1aI'm still getting agreement from our international partners about ELF,18:25
th1abut once we've agreed the governance structure we're going to put out18:25
th1athe call to get everyone contributing their expertise - the idea is18:25
th1athat as a community we identify service that we're interested in, and18:25
th1apropose tools, standards, and proto-standards (i.e. the REST interfaces18:25
th1awe've cooked up ourselves) for discussion, which the funding agencies18:25
th1a(in the UK, Australia and Canada initially) can then turn into official18:25
th1aguidance.18:25
jintyLooks like our spec would fit in there nicely18:27
jintyas a first draft18:27
th1aYes.18:28
jintywhat do they want to achieve?18:29
th1aInteroperability.18:29
th1aCETIS is useful in particular because they're, as an institution, somewhat skeptical of the super-complex IMS specs for eLearning.18:31
th1aWhich even I run in fear from.18:31
jintywhat are the IMS specs?? why run in fear18:31
th1aOK, so eLearning is all about doing online learning, which we aren't directly interested in, but it is closely related.18:32
th1aBut there is potential for immense amounts of money to be made from online delivery of educational content.18:33
th1aEspecially if it can be done in a modular, interoperable way.18:33
th1aa la carte, if you will.18:33
jintyTimetabling, from my perspective, doesn't seem to be content18:33
th1aSo that if I know I want to learn skill "foo," I can purchase modules that teach that exact thing from a number of companies and universities.18:34
th1ajinty:  we'll make it back to timetabling...18:34
jintySeems like something a lot of people are interested in, I saw somebody trying to do about the same.18:35
th1aSo... in an attempt to get that kind of interoperability, there are several big interlocking efforts.18:35
th1aThe military is also a player in this game, since they have complex training requirements.18:36
th1aAnyhow...18:36
th1aSo IMS is the biggest body of specs.18:36
jintyAs do all large organizations, and they have money.18:36
th1aAnd in particular they have "IMS Enterprise" which covers stuff which is directly related to SchoolTool.18:37
th1aSo IMS Enterprise has, for example, SOAP bindings for fetching a list of students in a class.18:37
jintyAh, penny drops. The administration system facilitates the delivery of content18:38
jintyadministering it, if you will18:38
th1aBut you're immediately stepping not only into SOAP, but this whole body of giant standards documents.18:38
th1aYeah, the administration system has a role.18:38
th1aFor example, in SchoolTool, if there is a certain requirement for a student to graduate that is defined as a standard,18:39
jintye.g. has to pass a certain combination of classes18:40
th1aand the student fulfills the requirement via an online class instead of a regular class, we'd like to receive that info from the elearning client in the standard manner.18:40
th1aParticularly if we can do so via REST, rather than SOAP.18:40
th1aSo basically, SchoolTool's point of view is that we want to support interoperability, but we want to promote simpler interoperability standards than IMS.18:41
th1aAnd its ilk.18:41
jintyAh, ok.18:42
th1aBut we'd also like someone else to someday add full IMS support.  It just isn't something Mark is likely to pay for.18:42
th1aSo the guys at CETIS share this point of view in general.18:42
th1aThey tend to point to things like RSS and iCalendar as more useful standards than the massive IMS type of things.18:43
jintyThey will or have generate their own standards?18:44
jintyAre we going to play a part in this?18:45
th1aWell, it seems like ELF is something we could play a part in.18:45
th1aSince, for example, they have some interest in timetabling standards.18:45
th1aAnd it would suck if they blessed something other than TTML.18:46
th1aNot that they have huge pull, but it would be a good indication of what way the wind is blowing.18:46
jintyTTML is more of an idea than a standard.18:46
jintyi.e. use MathML to define the constraints.18:47
th1aYeah.  It doesn't seem like anything else is floating around.18:47
jintythe MathML wind is very strong18:47
th1aI've only heard good things about it.18:48
jintywolfram, mathcad, almost all browsers are behind it.18:48
jintywolfram is the important one18:48
th1aActually, here's a good quote about ELF:  The intention is not to provide a blueprint for an open-source solution, but rather to facilitate the integration of commercial, home-grown, and open source components and applications within institutions and regional federations, by agreeing common service definitions, data models, and protocols.18:49
th1aHm.  Poking around... maybe there is an IMS Enterprise spec for timetabling.18:50
th1ahttp://www.brock.ac.uk/blogs/jon/PermaLink,guid,8d44008f-c9c3-4676-8e1b-fd95311f9407.aspx18:50
th1ahe says "i spent about 4 hours trying to understand and debug the IMS Enterprise Web Services WSDL files and schema yesterday. when i left the building (yes late) i felt like i'd given myself Asperger's syndrome"18:52
*** thisfred has quit IRC18:52
th1aThe objective here is to not make ourselves autistic.18:52
jintyhe he18:53
jintynot something I am interested in18:53
th1aI should probably email that guy.18:53
jintywhen they talk about timetabling data, what do they mean?18:54
th1aGood question.18:54
jintydoes it include the definition of constraints18:54
th1aActually it is unlikely that it does.18:54
jintyor is it just how to represent a timetable18:54
th1aIn fact, that is probably all they're talking about anyhow.18:55
th1aIn which case this entire thread of conversation is a dead end.18:55
th1aSo let's drop it for the moment.18:55
jintyIt's inetesting from the point of view of how to convert from this standard into something more suitable for a solver18:56
th1aLet's get back to what you've been writing.18:56
jintythat should be looked at now, so that the probems im solving now can take it into account18:56
th1aI'll investigate further.18:57
jintyThanks, if I can look at the data formats they use it would really help.18:57
th1aDo you know anything about SOAP?18:58
jintyNot a clue, I tend to avoid jargon18:58
th1aIt's the more heavy-weight way to do web services.18:58
th1aFavored by Microsoft, Sun, etc.18:59
jintynow define web services:)18:59
th1aTurns out that whenever someone like Amazon exposes the same services via REST and SOAP, 80-90% of developers use REST.18:59
th1aPassing XML over HTTP.18:59
th1ai.e., fetching an XML list of students in a class.19:00
th1aOr passing timetable constraints in TTML.19:00
th1aOver HTTP.19:00
jintyMy point with the timetabler is to try to isolate this code into the sources, so that the core doesn't have to care about how it gets the data19:01
th1aRight.19:01
jintyIt's the core I want to prove works right now.19:01
th1aRight.  Lets  get back to that.  I have to go in about 20 minutes.  I've got a doctor's appointment.19:02
jintyI don't want to get distracted by the other things becuse I know they are solvable problems. The greatest concentration of risk is the core19:02
th1aYes.19:03
jintyespecially the TTML part19:03
jintybasically how I see the core working, is that everything is stored in object form19:05
jintydirectly translatable to and from XML19:05
jintybecause XML can be completely expressed by objects19:05
th1aRight.19:06
jintyThis format is very general, and probably completely non-understandable to the average user19:06
jintybut, when you deal with a more specific problems, you can make some assumptions19:07
jintysuch as, a classroom scheduling problem will always only have 3 domains: Teachers, classes and time slots19:07
jintyusing these problem specific assumptions, you can come up with a usable GUI. I hope!19:08
th1aSure.19:08
th1aI'm getting more optimistic that it could be done well in a browser with some clever Javascript.19:09
jintyThats another thing I'm not thinking of - exactly what type of gui:)19:10
*** d2m has joined #schooltool19:10
th1aSo the biggest question at hand is whether or not our most fundamental modeling of the problem is adequate?19:11
jintyexactly. everything else is surely solvable and contains no risk.19:11
th1aAre you, in the end, interested in writing an optimization algorithm yourself?19:12
jintyI have been speaking to Tomaz (tablix) about it19:12
jintyHe was thinking of making TTML an input format. But the e-mail contact is patchy, so I don't know the situation19:13
jintyIt is very different from what he is doing right now19:13
th1aIs that all C?19:14
th1aTablix?19:14
jintyyep19:14
th1aI wonder if we could just wrap it in Python somehow, rather than writing a TTML parser in C.19:15
jintyanother option may be to write a tablix module that is generated from a TTML file. But this is a blue sky right now.19:15
th1aOK.  I think I understand where I should focus some attention.19:17
jintyFrom my understanding of them, a genetic algorithim only needs to be told when a constraint is violated and how badly. I am thinking of an addition to miniTTML that would do just that.19:17
jintyIt would also be usefull for letting the GUI know when a constraint is violated19:17
* jinty is not sure if he is making sense anymore19:18
th1aMore or less.19:19
th1aI need to run, however.19:19
jintyok, till later, I'll try to get some stuff on the ground when I have time19:20
th1aAll right.  I'll send out some inquiries about IMS Enterprise and timetabling.19:20
th1aGood news about that Zope Mag article.  Are you getting paid for that?19:21
jintyyep!19:21
th1aExcellent.19:21
jintyalso sounds like it could be a longer relationship19:22
th1aGreat.19:22
th1aOK, I have to find out whether I've got chronic low-grade bronchitis or chronic low-grade pneumonia.19:22
jintygood luck! At least both are low-grade19:23
th1aYeah.  I've just been coughing intermittently for two months.19:23
th1attyl.19:23
jintyouch!19:23
*** alga has joined #SchoolTool19:25
*** th1a has quit IRC19:41
*** hazmat has joined #schooltool19:53
*** hazmat_ has joined #schooltool20:02
*** hazmat has quit IRC20:03
*** faassen has left #schooltool20:08
*** Aiste has quit IRC20:08
*** Aiste has joined #schooltool20:19
*** bskahan has joined #schooltool21:08
*** jinty has quit IRC21:14
*** SteveA has quit IRC21:18
*** bskahan has quit IRC21:53
*** hazmat_ is now known as hazmat22:09
*** th1a has joined #schooltool22:40
*** mgedmin has quit IRC23:17
*** alga has quit IRC23:34

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