*** menesis has quit IRC | 02:44 | |
*** yvl has joined #schooltool | 09:45 | |
*** menesis has joined #schooltool | 10:21 | |
*** menesis has quit IRC | 13:11 | |
*** menesis has joined #schooltool | 14:30 | |
*** replaceafill has joined #schooltool | 15:07 | |
*** th1a has joined #schooltool | 15:31 | |
th1a | OK... had a lunch pickup snafu. Give me 10 minutes... | 15:31 |
---|---|---|
yvl | ok | 15:34 |
th1a | All right. Our regular ride over to nursery had their car in the garage. | 15:38 |
th1a | hi yvl, menesis, replaceafill. | 15:38 |
replaceafill | good afternoon | 15:39 |
yvl | hi | 15:39 |
menesis | hi | 15:39 |
th1a | I looked at the relationship branch yvl. Seemed to work. | 15:40 |
yvl | kk | 15:40 |
th1a | The UI seems good. | 15:41 |
th1a | Anything to report yvl? | 15:43 |
yvl | yes | 15:43 |
yvl | ok | 15:43 |
yvl | so I worked this weekend as I feel I'm running out of time :) | 15:44 |
th1a | Yes... it is getting a bit more tangible. | 15:44 |
yvl | I wrote an alternative optimization along with an evolution script | 15:44 |
yvl | I'm testing it against ACC database | 15:45 |
yvl | failed to finish today | 15:45 |
yvl | I'll give you a spreadsheet with page load times | 15:45 |
yvl | and data that needs to be deserialized on each page | 15:45 |
yvl | school tab, students group, searching people, random CanDo gradebook, random person and searching for skills | 15:46 |
yvl | I'm testing those against three versions of ST: | 15:46 |
yvl | 1) current trunk | 15:46 |
yvl | 2) temporal relationships with some optimizations (that you have tested) | 15:47 |
yvl | 3) temporal relationships with global catalogs | 15:47 |
yvl | tests are quite simple | 15:47 |
yvl | start ST | 15:47 |
yvl | load a page on "cold" database | 15:47 |
yvl | load the page again | 15:47 |
yvl | for example, school tab on current trunk on my machine: | 15:48 |
yvl | cold - 40 secs, hot - 10 secs | 15:48 |
yvl | with temporal and weekend optimizations: | 15:48 |
yvl | cold - 6.1 secs, hot - 0.6 secs | 15:48 |
yvl | note that temporal relationships add some of the slowness | 15:49 |
yvl | in hot school tab they account for at least 0.3 secs | 15:49 |
yvl | it was tough, but I think it's was worthwhile | 15:50 |
yvl | so I plan to spend most of the tomorrow testing | 15:50 |
th1a | Yes, definitely. | 15:50 |
th1a | Have our ACC users just accepted slowness or is it actually faster on their machine. | 15:51 |
* yvl shrugs | 15:51 | |
yvl | I do hope it runs faster | 15:51 |
th1a | I guess most common views would be "hot" in production. | 15:51 |
replaceafill | it takes around 5 seconds to load | 15:51 |
replaceafill | hot ofc | 15:51 |
th1a | How much memory on that thing? | 15:52 |
replaceafill | hold on | 15:52 |
replaceafill | Mem: 3069440 2878572 190868 0 148216 1154872 | 15:52 |
replaceafill | oops | 15:52 |
replaceafill | 3 GB | 15:52 |
th1a | How many students? | 15:52 |
replaceafill | 11.6k | 15:53 |
th1a | Oh. OK. | 15:53 |
th1a | So... way, way more than one school. | 15:53 |
th1a | That's a relief. | 15:54 |
yvl | they have buildings for locations | 15:54 |
th1a | Right. | 15:54 |
th1a | Yes. | 15:54 |
yvl | 2000 teachers | 15:54 |
yvl | 256 sections | 15:54 |
yvl | 40.000 skills | 15:55 |
replaceafill | yvl, ACC? | 15:55 |
replaceafill | ah | 15:55 |
yvl | I think so | 15:55 |
replaceafill | you mean socialskills | 15:55 |
replaceafill | that's the autism center | 15:55 |
yvl | no | 15:55 |
yvl | APS | 15:55 |
replaceafill | weird, i only see 852 teachers in the production server | 15:56 |
yvl | by no I meant "wait a minute" :) | 15:56 |
yvl | Data.fs-2012-11-09_14-44-32 | 15:56 |
replaceafill | right, that's autism | 15:56 |
* replaceafill is confused now :D | 15:57 | |
replaceafill | i should use prefixes for backups ;) | 15:57 |
yvl | :D | 15:58 |
yvl | true | 15:58 |
replaceafill | continue yvl, sorry for interrupting | 15:58 |
yvl | it's ok, I'm pretty much done | 15:58 |
yvl | after testing, we'll have to decide which approach we are taking | 15:58 |
yvl | well, it will hopefully be pretty obvious | 15:59 |
th1a | Does one approach have a lot of necessary follow-up tasks? | 15:59 |
yvl | no | 16:00 |
yvl | but the one I did this weekend involves a trickier evolution | 16:00 |
yvl | but it's been written | 16:00 |
yvl | and seems to work on APS database | 16:00 |
th1a | OK, so pretty much just see what's fastest? | 16:01 |
yvl | yep | 16:01 |
yvl | or rather "is the second one much faster" | 16:01 |
yvl | because if gains are negligible, I'd slightly lean towards the first one | 16:02 |
yvl | because it's evolution is simpler | 16:02 |
yvl | so I trust it more :) | 16:02 |
yvl | but just a little bit | 16:02 |
th1a | So if it is the same, you'd prefer without catalogs? | 16:03 |
yvl | yes | 16:03 |
th1a | OK. Well, we'll see how the numbers look. | 16:04 |
yvl | but overall I'd prefer the faster one | 16:04 |
th1a | Yes! | 16:04 |
th1a | I'm glad you did this. | 16:04 |
yvl | thanks! | 16:04 |
* yvl too | 16:05 | |
th1a | does one or the other make a difference in memory consumption? | 16:05 |
yvl | yes | 16:06 |
yvl | it's a bit tough to test exactly | 16:07 |
th1a | Hm... | 16:07 |
yvl | because what matters to us is memory consumption after day's usage | 16:07 |
yvl | and we haven't written any load testing scripts | 16:07 |
yvl | from a glance, it seems that we use less memory with optimizations | 16:08 |
yvl | but | 16:08 |
yvl | since optimizations are "do not load extra data unless it's really really needed" | 16:08 |
yvl | the memory consumption may be less, or greater, or roughly equal after heavy usage | 16:09 |
yvl | instinctively I'd say "somewhat less" | 16:09 |
th1a | OK, I see. | 16:09 |
th1a | So avoiding loading things you don't need in a given case doesn't mean much over time. | 16:10 |
yvl | maybe yes, maybe no | 16:10 |
th1a | Doesn't necessarily. | 16:10 |
yvl | yes | 16:10 |
yvl | because maybe it will not load these things at all | 16:10 |
yvl | maybe it will just keep the important btrees hot | 16:11 |
yvl | and flush everything else out | 16:11 |
th1a | OK. | 16:12 |
th1a | Thanks yvl. | 16:12 |
th1a | replaceafill? | 16:12 |
replaceafill | ok | 16:12 |
replaceafill | i caught a cold last week, so i worked during the weekend too :) | 16:13 |
replaceafill | http://dev.schooltool.org:7080/manage | 16:13 |
replaceafill | manager:schooltool | 16:13 |
th1a | I've got a cold now... | 16:13 |
replaceafill | :( | 16:13 |
replaceafill | go to Courses | 16:13 |
replaceafill | and add a Course | 16:13 |
replaceafill | you'll see a Level attribute in the add view | 16:14 |
replaceafill | the attribute also appears in the index view of the course | 16:14 |
replaceafill | and in the edit view | 16:14 |
replaceafill | th1a, question | 16:14 |
replaceafill | is this how you want to handle it? | 16:14 |
th1a | How else would we handle it? | 16:14 |
replaceafill | i mean, aren't we going to include courses information in the level pages | 16:15 |
replaceafill | btw | 16:15 |
replaceafill | levels only have titles right now | 16:15 |
replaceafill | i remembered your comments about standard metadata, etc :) | 16:15 |
th1a | Hrm. | 16:15 |
replaceafill | not sure if descriptions do something for levels | 16:15 |
replaceafill | i mean, if we should add them | 16:16 |
th1a | I have a hard time imagining anything else is necessary. | 16:16 |
replaceafill | kk | 16:16 |
th1a | I mean, level "9" = "Freshman" | 16:16 |
th1a | But that's not really used as a description or anything. | 16:16 |
replaceafill | right | 16:16 |
replaceafill | so, i was thinking maybe we should list the courses associated with a level | 16:17 |
replaceafill | in the level view? | 16:17 |
th1a | We could but it doesn't strike me as particularly important. | 16:17 |
replaceafill | ah ok | 16:17 |
replaceafill | never mind then :) | 16:17 |
replaceafill | so, this add/view/edit in course is enough? | 16:17 |
th1a | Yeah. | 16:18 |
replaceafill | ah ok | 16:18 |
th1a | It isn't rocket science. | 16:18 |
replaceafill | i was expecting more changes | 16:18 |
replaceafill | importers? | 16:18 |
th1a | If you move the courses to the next year, that'll need to be included. | 16:18 |
th1a | Yes, importers. | 16:18 |
replaceafill | by move you mean the option to copy courses from one year to the new one, correct? | 16:19 |
th1a | Yeah. | 16:19 |
replaceafill | so if math for level 2 is copied, should it be math for level 3? | 16:20 |
replaceafill | in the new year? | 16:20 |
th1a | No. | 16:20 |
th1a | But we need to think about adding that kind of logic for moving STUDENTS. | 16:21 |
replaceafill | that would be the "promote students" feature, right? | 16:21 |
replaceafill | so, courses should just copy the level data from the previous year | 16:21 |
th1a | Yes, promoting students is more complicated. | 16:22 |
th1a | Students aren't assigned levels at all right now, correct? | 16:23 |
replaceafill | no | 16:23 |
replaceafill | we'd need to use the section <-> course <-> level relationships | 16:23 |
th1a | OK. So that probably needs to be "temporal." | 16:23 |
replaceafill | ah | 16:23 |
th1a | But it also will mostly be done as a batch job between years. | 16:24 |
th1a | If that makes sense. | 16:24 |
replaceafill | yes | 16:24 |
replaceafill | should we add level information to the importer? | 16:25 |
replaceafill | i mean, a Levels sheet? | 16:25 |
th1a | Oh, yes. | 16:25 |
replaceafill | ok | 16:25 |
th1a | And a column to courses, etc. | 16:25 |
replaceafill | i'll work on that next | 16:25 |
replaceafill | sure | 16:25 |
replaceafill | i have a quick question for menesis | 16:25 |
replaceafill | menesis, this instructions won't work in 2.6, right? | 16:26 |
replaceafill | http://book.schooltool.org/multiple-schooltools.html | 16:26 |
replaceafill | (a user asked about this in spanish) | 16:26 |
replaceafill | (https://answers.launchpad.net/schooltool/+question/238437) | 16:26 |
menesis | replaceafill: in general, this would work | 16:27 |
replaceafill | ah | 16:27 |
menesis | but yes, there are more files that need to be modified now | 16:27 |
replaceafill | i thought that people would need to at least change the redis port | 16:27 |
replaceafill | ah ok | 16:27 |
replaceafill | we'll see if the user comes back with a new question :) | 16:28 |
* replaceafill done | 16:28 | |
menesis | zeo.conf needs to point to a new Data.fs, schooltool.conf has different database definition, supervisord.conf needs many changes | 16:29 |
th1a | It seems like there have been a lot of questions in spanish or maybe it is the same thread. | 16:29 |
replaceafill | there are users helping users now :) | 16:29 |
menesis | :) | 16:30 |
th1a | Ah, in Spanish? | 16:30 |
replaceafill | yes | 16:30 |
th1a | Cool. | 16:31 |
th1a | OK, thanks replaceafill. | 16:31 |
th1a | None of these levels related things should be individually complicated. | 16:31 |
th1a | It is just wiring all the parts together. | 16:31 |
th1a | menesis? | 16:31 |
menesis | I was on vacation since Monday | 16:32 |
th1a | Yes. | 16:33 |
th1a | Refreshing? | 16:33 |
menesis | have not been reading email even. | 16:33 |
th1a | OK. | 16:33 |
menesis | well. a break is good. | 16:33 |
th1a | Sure. | 16:34 |
th1a | OK. | 16:34 |
th1a | Are we done then? | 16:35 |
menesis | yes | 16:35 |
th1a | OK. | 16:36 |
th1a | Thanks guys. | 16:36 |
th1a | Hopefully I'll shake this cold quickly... | 16:36 |
th1a | Have a good week. See you Wednesday. | 16:36 |
replaceafill | thanks everybody | 16:36 |
* th1a drops the bag of gravel. | 16:37 | |
yvl | thanks guys | 16:39 |
*** yvl has quit IRC | 16:54 | |
replaceafill | th1a, the right term is "Grade Level", correct? | 17:27 |
replaceafill | (based on the sidebar link in /manage) | 17:28 |
th1a | replaceafill: yes. | 18:31 |
replaceafill | th1a, kk | 18:31 |
*** menesis has quit IRC | 18:51 | |
*** replaceafill has quit IRC | 21:03 | |
*** menesis has joined #schooltool | 21:21 | |
*** replaceafill has joined #schooltool | 21:58 | |
*** menesis has quit IRC | 23:43 | |
*** menesis has joined #schooltool | 23:44 | |
*** menesis has quit IRC | 23:51 | |
*** menesis has joined #schooltool | 23:54 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!