IRC log of #schooltool for Wednesday, 2012-11-07

*** khildin has joined #schooltool00:08
*** replaceafill has quit IRC00:27
*** menesis has quit IRC00:48
*** khildin has quit IRC01:47
*** ignas has quit IRC03:22
*** vialde has quit IRC03:43
*** ignas has joined #schooltool04:53
*** ignas has quit IRC07:10
*** khildin has joined #schooltool08:47
*** yvl has joined #schooltool10:22
*** paulproteus has quit IRC10:32
*** paulproteus has joined #schooltool10:42
*** vialde has joined #schooltool11:10
*** menesis has joined #schooltool11:40
*** vialde has quit IRC12:25
*** vialde has joined #schooltool12:26
*** ignas has joined #schooltool13:05
*** menesis has quit IRC14:54
*** menesis has joined #schooltool16:23
th1a hi yvl, menesis, aelkner.16:30
veloutingood morning16:31
yvlhi guys16:31
th1aGlad to be out from under that election.16:32
th1aWith the right result.  ;-)16:32
yvlcongratulations with a new president :)16:32
th1aSame as the old one.16:32
yvl+1 :)16:32
th1aHow's our fix looking yvl?16:33
yvlabout to be done16:33
yvlit was quite fun actually16:34
yvlwhat's done if they apply the new code to an older database, scores will not be broken16:34
yvlwhat's about to be done - when they apply new code to new database + modified scores in some sections, old scores will reappear and not kill the fresh ones16:35
th1aRight.  Good.16:35
yvland I actually found one more bug while checking today16:36
yvlthank goodness I had the Data.fs16:36
yvlbecause when you're reading the code, it's almost impossible to spot variables that are named very similarly but wrong16:37
yvleven with Data.fs I spent some half our looking and thinking what the?16:37
yvllooking good16:37
yvlwill finish today I think16:38
yvl(during the meeting)16:38
th1aI think replaceafill can test against dwelsh's instances.16:40
th1aShould we plan on pushing a .deb fix tomorrow?16:40
yvlsounds good to me16:41
menesisbut you want to test it first?16:41
aelkneralso, do we want it to include the search changes?16:41
th1aWe can at least test it on dwelsh's data today.16:42
th1aI don't want to rush the search changes.16:42
th1aLet's just keep this a bugfix.16:42
th1aSo we know this works with Stephen's Data.fs, right?16:43
th1a(to confirm the obvious)16:43
th1ayvl:  Is there any reason some people would have this bug and not others?   Terms?16:43
yvlso, yes, this works with Stephen's Data.fs16:44
yvlhis scenario that triggered the bug was16:44
yvlthat they created new sections of courses16:45
yvland those sections had same instructors16:45
yvland they upgraded at that point16:45
yvlthey were setting up spring 2013 sections it seems16:46
yvlso that's the scenario that breaks stuff16:47
yvlI would say it's a common one16:47
th1aDid you see the traceback I just forwarded?16:47
yvlyes, but didn't look closely16:48
yvlI think I saw that before somewhere16:48
yvlbut can't recall it16:49
yvllooks a bit esoteric16:49
th1aOK.  If it survives into tomorrow, you may have to field it.16:49
th1aI'll let you know.16:49
th1aOtherwise, I guess you can get back to reports.16:50
th1aThanks yvl.16:51
yvlone more thing16:51
aelknerafter yvl looks into the node index while testing issue, right?16:51
yvlData.fs'es are always helpful in such cases, th1a16:51
yvldon't want to push it, but if they are in a mood to share... :)16:52
yvlbtw, aelkner16:52
yvldo any of CanDo imports create/assign/modify persons or their groups?16:53
aelknerschooltool.virginia has a special section importer16:54
aelkneri think it assigns groups, but i have to check16:54
yvlit seems that some modified events were not fired and some of the objects were not indexed16:55
yvland it is related to groups, persons or more likely their group membership16:55
th1aDo you want to take first crack at that aelkner?16:55
yvlif it keeps broken, please point me to a branch of schooltool.virigina they have deployed16:56
aelkneryvl, i don't think it has to do with schooltool.virginia16:58
yvlso I don't know what they have deployed16:58
yvlbut first place to look is at whatever imports they were dealing with16:58
aelknerwait, are we talking about the broken tests and the updateIndexes call?16:58
yvlthat touched groups, persons, section membership and such16:58
yvlno aelkner16:59
yvltraceback th1a sent16:59
aelkneri just checked and schooltool.virginia importer doesn't touch group membership17:00
aelknerjust section membership17:00
aelkneralso, section courses and resources get updated17:01
th1aOK, well, let's table this for a minute.17:01
th1aSo... we've got this "node index while testing issue?"   yvl?17:02
yvlyes, the one we talked Monday17:03
yvlfix test set-up17:03
yvlI got aelkner's email -- thanks btw!17:03
yvlso I'll fix that once other things are done17:04
aelkneryou asked, and you shall receive :)17:04
th1aAnything else, yvl?17:04
yvlother things being only that traceback17:04
yvlI still want to be pointed to the code they have deployed :)17:04
yvlif they use schooltool.virginia - which revision or version or such and where I can find it :)17:05
yvl(assuming you want me to look into that traceback tomorrow)17:05
th1aEither aelkner or replaceafill should be able to have a first crack at it.17:06
th1aThanks yvl.17:07
menesisI have read juju docs and written a simple charm17:08
menesisalso looked at some zope changes and what needs upgrading17:10
th1aI've been wondering about how one adds their own charms.17:10
menesiswas trying to do that now17:12
menesisjuju deploy --repository=. local:precise/schooltool17:14
menesisprecise/schooltool is a directory where the charm is17:15
th1aOK, well, let me know when you have something I can try.17:15
menesisbut I am still waiting for it to do something17:16
th1aAs in?17:16
menesisdid not get an install to work yet17:17
menesiswill tell when have one running17:17
menesisnothing interesting to say atm17:18
th1aAll right.  And shooting for a bugfix release tomorrow.17:18
th1aOK, thanks menesis.17:19
aelknerok, i changed the search view filter to return no items when view is first called up17:20
aelknerthen i added the No layers assigned checkbox and added the filter logic for it17:20
aelkneri fixed the broken tests that resulted from the changes, and sent word to yvl, as we discussed17:22
aelknerin so doing, i noticed one more thing about node layer filtering17:22
aelknerthe import_skillls.xls test that i wrote a while ago was based on the original unit tests17:23
aelknerso some of the layers are not really rational, but they are theoretically possible17:24
aelknerso they still come up in the search17:25
aelkneralso, the checkboxes are based on the document layers17:25
aelknerso id a user adds a node with a layer that is not in a document, then it will not be findable17:26
aelknerunless we added yet another checkbox, 'nodes with weird layers'17:26
aelkneronly kidding, ofc17:26
aelkneralthough i am seriously asking what you want to do about that case17:27
th1aThere is no reason in particular not to list all layers in the system, right?17:27
aelknerah, sorry, that wasn't it, i just realized something17:28
aelkneryes, all layers are listed as checkboxes, scratch what i just said about that17:28
aelknerthe issue was this:17:28
aelknerin my silly test, again theoretically possible, a node was added17:28
aelknerand the layer was a skillset layer17:28
aelknerskillsets layers are grouped with the skillset checkbox17:29
aelknerso the current logic can never return those nodes17:29
th1aBecause they are nodes and not skillsets?17:30
th1aMaybe you should just prevent those from being created.17:30
aelknerand since the skillset layers are identified as such, they are removed from the nodes layer-based search17:31
aelkneri guess you mean have a check similar to cyclic relationship that throws an exception17:32
aelknerif a layer is chosen that is a skill or skillset layer17:32
th1aI suppose so.17:33
aelknerthing is, it's another place where the very idea of what is a skillset or skill layer is determined17:33
aelknerwhen the document heirarchy is set17:33
aelknerso, in theory, a layer could be added to a node that later turns out to be a skillset layer17:34
aelknertoo late :(17:34
th1aOK, so basically, it is a bug.17:35
aelknerwell, it's more like a situation that calls for a design decision17:36
aelkneri think we need to anticipate such weird nodes being created17:36
aelknerand have a way to find them17:36
aelkneras long as we allow the user to poke around nodes and change layers and such17:37
aelknerthey have to be able to find nodes they may have broken17:37
th1aI'm pretty certain that when I first explained this requirement to you I suggested we just give documents simple attributes to remember aliases for "skillset" and "skill" but you didn't listen.17:37
th1aAnything other than that seems like a kludge.17:38
aelknerwell, remember, the use case was to have a hierarchy of labels for adding the nodes down to the skills17:39
th1aBut basically, I'm willing to just chalk these weird cases as a medium level bug and forget about it.17:39
aelknerand layers was the object that we had available for that purpose17:39
th1aThe use case was always just to have different names for skillsets and skills in the context of the document.17:40
th1aThat is all we ever wanted or needed.17:40
aelkneryeah, well, considering that it is only a theoretical case17:40
aelknerand using the document views will avoid it17:40
aelkneri can see chalking it as you suggest17:40
th1aI'm very much ready to put this behind us.17:41
aelknerregarding the design decision to use layers for skillsets and skills17:41
aelkneri'm surprised to hear raise an objection to that now17:42
aelknerit's not lke you haven't had two months since i did it to say something17:42
aelkneryou know, restating a point until it is understood is not against the law17:42
th1aI wouldn't be complaining if it worked well.17:43
aelknertbw, i would have loved to have not used layers, i just didn't see how17:43
aelkneri suppose we could have had the two special attributes as you suggest17:43
th1aOK, so when can we put this behind us?17:44
aelknerwe kinda have to live with what we have regarding layers for skillsets and skills17:44
aelknerand yes, we can put behind us the issue with the funny nodes17:44
aelknerthey are unlikely to happen in the real world, just possible in theory17:45
aelknerso, still to be done:17:45
aelkner1) awaiting the indexing fix from yvl, then i'll incorporate it17:45
aelkner2) i need to remove the skillset checkboxes from nodes-only search (edit child nodes view)17:46
aelkner3) i need new tests, i only fixed existing ones, but we should have search tests that manipulate the checkboxes17:46
aelknerno more than a day, day and half or that17:47
aelknerall three that is17:47
aelknerwe should try to get that into a release17:48
aelknerso welsh can have it?17:48
aelknerdev ppa, ofc17:48
th1aCan you start by triaging that import traceback today... try to get a handle on what its source is?17:48
th1aWe'll get it out, but not immediately.17:49
aelkneri'm still wondering what you guys are referring to there17:49
aelknerwas there a launchpad bug added17:49
aelkneror an email that i inadvertently deleted?17:49
th1aFwd: CanDo Follow-up17:50
aelknerah, the new one17:51
aelknerjust noticed it17:52
aelknerso i see an id found in an index is corrupt, that's what yvl was talking about17:54
aelknerapparently the importer is suspected of corrupting an index?!17:54
aelkneri looked back at the chat and yvl asked if any of the importers change person group membership17:56
aelknerto answer the question in a more focused manner:17:56
aelknerschooltool.cando importers, no17:56
aelknerschooltool.virginia importer changes person section membership and section courses and resources17:57
aelknerschooltool core importer does change person group membership in various sheets17:57
aelknerbut that's by design17:57
aelknerbut i don't see how that would break an index17:58
th1aWell, take a look at it.18:00
th1aI'm going to wrap this up...18:00
th1aThanks guys.18:00
th1aHave a good week and weekend.18:01
* th1a drops the bag of gravel.18:01
aelknerok, thanks guys, great weekend18:01
yvlthanks guys18:07
yvlok, th1a fix committed18:08
th1aThanks yvl.18:20
th1aaelkner:  ayt?18:55
aelknerth1a, yes?18:57
th1aI forwarded a bug from Welsh to you just now.18:57
th1aSince replaceafill is MIA.18:57
aelknerthese tracebacks make me nervous, i mean how can one guess why they happen19:00
th1aWell, I guess replaceafill has access to the instance.19:02
aelkneryes, that always helps19:02
aelknerone needs to be able to match a version of codebase with a data file and recreate the problem on one's machine19:04
aelknerbut you know what, i could make one guess19:04
aelknercould this have to do with the importer blowing away the skill equivalencies for project skills?19:05
aelkneri believe replaceafill coded a change to that import, but i doubt welsh has it19:06
aelkneralso, even if he did, he would need to revert to the data.fs that was pre-reimport, if that's what happened19:06
th1aSo basically we should wait for replaceafill to appear?19:07
aelkneryou could ask menesis if he knows of the import change and if he can release it19:08
th1aWell, we're doing a release tomorrow.19:08
aelknerif that's commited, pushed yet or not19:08
aelknerperhaps it includes the import fix?19:08
th1aYes, that will need to be in the release.19:08
aelknerok, that's good, i would tell welsh that he needs to revert to pre-reimport data.fs19:09
aelknerand wait for the new release before reimporting19:09
aelknerif that is in fact what happened19:10
th1aI'll wait to see if replaceafill is alive since he can do whatever is necessary anyhow.19:10
aelknerok, regarding the other traceback, who is the customer on that one?19:11
aelkneroh, glenda, right?19:11
aelkneryou know what, it could be the same reimport problem19:12
aelknerwith the equivalcies19:12
aelknerbecause she says the gradebooks are broken19:12
aelknerbut the traceback with the intid crash in the table formatter is not related to equivalencies19:13
aelknerthat has something to do with group membership, a total mistery to me19:14
aelkneri don't see what the importing would do to break a intid in an index19:15
aelknerall the code does is do things like: students.remove(student) and students.add(student)19:16
aelknerthat shouldn't break any index or item in an index19:16
aelkneryvl should probably look at that one, perhaps there's some tricky gotcha in schooltool.relationship?19:17
aelknerbut i suspect you'll need to advise glenda to revert to pre-reimport data.fs as well19:21
th1aI think this is a fresh install anyhow.19:21
aelknerand to make sure that after restoring the data.fs, the problems go away19:21
th1aI'll send a follow-up.19:22
*** ignas has quit IRC19:34
*** replaceafill has joined #schooltool20:13
replaceafillth1a, zyt?20:13
th1ahi replaceafill.20:19
replaceafillhey th1a20:19
replaceafilli'm not MIA >:(20:19
replaceafilli'm traveling ;)20:19
replaceafilli'm in houston right now20:19
th1aoh yeah!20:19
replaceafillabout to get into my next flight20:19
replaceafillbut i wanted to ask you20:19
replaceafilli'm seeing a lot of tracebacks and errors20:19
replaceafilland fixes20:19
replaceafilli think i'll do some work when i get to the hotel20:20
replaceafilldo you need me to do anything specific?20:20
th1aMainly look at dwelsh's since you've got access to the server, I think.20:20
th1ado that20:20
replaceafillthe traceback he reported today, right?20:21
replaceafillkk, will do20:21
*** replaceafill has quit IRC20:36
*** paulproteus has quit IRC20:39
*** menesis has quit IRC20:46
*** paulproteus has joined #schooltool20:47
*** OTNexus has joined #schooltool22:54
OTNexusne1 online?22:54
th1ahi OTNexus.22:56
OTNexushey :)22:56
th1aCan I help you with something?22:58
*** OTNexus has quit IRC23:24

Generated by 2.15.1 by Marius Gedminas - find it at!