IRC log of #schooltool for Monday, 2005-09-19

*** povbot has joined #schooltool00:02
*** Aiste has quit IRC00:15
*** yvesC has quit IRC01:37
*** maharaja has quit IRC01:43
*** maharaja has joined #schooltool02:23
*** th1a has quit IRC03:59
*** th1a has joined #schooltool04:00
*** maharaja has quit IRC06:32
*** tvon changes topic to "SchoolTool development | IRC logs at http://source.schooltool.org/irclogs/ | Buildbot: http://source.schooltool.org/buildbot/ | Dev meetings Mon, 13:30 UTC | CanDo dev meetings Tue, 4pm EST"07:42
*** strichter has joined #schooltool08:46
*** srichter has quit IRC08:46
*** _srichter has joined #schooltool08:54
*** strichter has quit IRC08:54
*** mgedmin has joined #schooltool11:15
*** alga has joined #SchoolTool12:04
*** Aiste has joined #schooltool12:17
*** _srichter is now known as srichter13:28
povbot/svn/commits: * mg committed revision 5006:14:34
povbot/svn/commits: Bugfix for issue379: PDF calendar views always use UTC.14:34
povbot/svn/commits: I haven't functionally tested this.14:34
*** thisfred has joined #schooltool14:38
povbot/svn/commits: * mg committed revision 5007:14:42
povbot/svn/commits: Timezone cleanups in schoolbell.calendar:14:42
povbot/svn/commits: - Require `first` and `last` arguments to expand to have explicit timezones14:42
povbot/svn/commits: (instead of falling back to UTC).14:42
povbot/svn/commits: - Do not insist that `first` and `last` have the same timezone.14:42
povbot/svn/commits: - Changed parse_datetimetz to accept ISO 8601 dates in UTC (i.e. with a 'Z' at14:42
povbot/svn/commits: the end).14:42
povbot/svn/commits: - Moved argument checks from CalendarEventMixin.expand to CalendarMixin.expand14:42
povbot/svn/commits: (optimisation; about 10-15% speedup in benchmark/nonrecurrent.py).14:42
povbot/svn/commits: - Removed timezones and all-day events from "Future goals" section in the README14:43
povbot/svn/commits: (they're implemented, although I have some suspicions as to the correctness of the implementation).14:43
povbot/svn/commits: - Fixed a corner case description in the CalendarMixin.expand doctest.14:43
*** jinty has joined #schooltool14:47
povbot/svn/commits: * srichter committed revision 5008:14:54
povbot/svn/commits: Added update target and abstracted the checkout stuff a bit more, so that it is easier to switch to writable checkouts of the Zope code (which I do all the time).14:54
mgedminsrichter, svn co svn://svn.zope.org/repos/main//branches/testbrowser-integration/src/zope/testbrowser Zope3/src/zope/testbrowser14:56
mgedminsvn: URL 'svn://svn.zope.org/repos/main/branches/testbrowser-integration/src/zope/testbrowser' doesn't exist14:56
mgedminmake: [testbrowser-checkout] Error 1 (ignored)14:56
mgedminACL view tests fail14:57
mgedmina problem with macros?14:58
srichterwell, everything fails because of the new Zoep 3 version14:58
srichterI am fixing the pagelet stuff right now14:58
mgedminah, I see14:58
povbot/svn/commits: * jinty committed revision 5009:14:58
povbot/svn/commits: Change private svn+ssh urls pointing at srichter refactor, to public ones pointing at trunk.14:58
mgedminpreviously everything failed because Zope 3 was too old14:59
mgedminnow everything fails because Zope 3 is too new14:59
mgedmintracking the trunk is sure fun :-)14:59
jintyhehe14:59
srichteryep15:00
srichterbut it was my development that broke it, so it si controlled :-)15:00
jintycreative destruction15:00
*** bskahan has joined #schooltool15:03
bskahanmgedmin, why are you opposed to changing the CSS style?15:03
bskahan#NavigationColumn DIV.Portlet H5 {} is much more readable than #navigation_column div.portlet h5 {}15:05
mgedminbskahan, HTML tags in uppercase rub me the wrong way15:05
mgedminno it isn't, not to me15:06
bskahanits CSS, not XML15:06
mgedminXHTML is case sensitive, I believe15:06
bskahanthis is css, not xhtml15:06
mgedminreferring to a case-sensitive lowercase tag via a uppercase name in CSS, even though it works, looks just plain wrong to me15:06
mgedminpersonal preference15:06
* mgedmin also hates literal TAB characters15:07
bskahanparticularly, .body {} vs BODY {}15:07
srichtermgedmin: this is pretty arbitrary15:07
bskahanin this case there's a reason to use tabs vs. spaces, 1 char weight vs. 4 or 815:07
mgedminsrichter, most coding conventions are15:08
srichterbut most coding conventions we use at least are documented and agreed upon by a group of developers15:08
bskahanupper case tag names is not uncommon, but there is no pep-8 for CSS style15:09
* mgedmin nods15:09
mgedminCSS2 spec:15:10
mgedmin"The case-sensitivity of document language element names in selectors depends on the document language. For example, in HTML, element names are case-insensitive, but in XML they are case-sensitive."15:10
mgedminwhat about XHTML? XHTML is XML, isn't it?15:10
mgedmintherefore case-sensitive15:10
mgedmincase closed15:10
* mgedmin is being a bastard today15:10
* mgedmin needs feeeding15:10
bskahanif its in the CSS2 spec then I apologize15:11
mgedminhttp://www.w3.org/TR/REC-CSS2/selector.html, section 5.1, second paragraph15:11
bskahanxhtml is still served as text/html which is probably why it still works15:11
bskahanthat's too bad, I still think it more readable15:12
bskahanI'll change it back15:12
bskahancan we agree on CamelCase ids and cross page classes, and lower_case single page classes?15:15
* mgedmin doesn't object15:15
srichtercan we always use CamelCase? :-)15:15
srichterswitching between notations is confusing15:15
bskahanI'd like to distinguish between major classes that are part of the system style and section specific style that only applies to one view (daily calendar for ex.)15:16
bskahanwhat about a section prefix?15:18
bskahanCalDailyFoo?15:18
srichterthat's fine with me15:19
bskahanmgedmin?15:19
povbot/svn/commits: * srichter committed revision 5010:15:22
povbot/svn/commits: Switch to latest viewlet code. This fixes all ftest failures, but unit tests are still failing. People render page templates in unit tests, sigh.  :-(15:22
mgedminbskahan, +015:22
mgedminactually, +0.515:22
mgedminsome convention is better than no convention at all15:22
mgedminsrichter, you have discovered the secret of SchoolTool: we render page templates in unit tests15:23
mgedminin hindsight, perhaps that wasn't one of the best ideas15:23
srichterno kidding15:24
*** Aiste has quit IRC15:24
srichterso now I am contemplating whether I will write a huge setup to support the setup of viewlets in unit tests or change the tests15:24
jintysrichter: a small patch for zpkgtools http://paste.plone.org/3944. reference: http://issues.schooltool.org/issue17015:25
mgedminsrichter, isn't that like five lines setting up adapters/views in ViewTestBase.setUp?15:25
srichterwell, you need to make sure that all viewlet regions that could ever be called are available15:26
mgedminadd them one by one, when you encounter a failing test15:26
srichterI really would hope that I do not need to register all viewlets, since they are often not interesting to the test15:26
srichterbut this produces brittle tests15:27
srichterand ST is full of them, so I don't want to add more15:27
bskahanare there many places that we render content in unit tests?15:27
mgedminyes15:28
bskahanI know of some, just wasn't sure how man15:28
bskahany15:28
srichterI have 20 failures and 4 errors15:28
srichterjinty: send the patch to Fred15:29
jintyok15:29
*** mgedmin has quit IRC15:30
povbot/svn/commits: * srichter committed revision 5011:15:36
povbot/svn/commits: Of course we have to point in our site.zcml to the viewlet package as well.15:36
povbot/svn/commits: * jinty committed revision 5012:15:50
povbot/svn/commits: Fail on error when compiling translations.15:50
*** mgedmin has joined #schooltool15:50
*** mgedmin is now known as mgedmin|gprs15:51
*** FarcePest has joined #schooltool16:01
povbot/svn/commits: * srichter committed revision 5013:16:08
povbot/svn/commits: Ok, the unit tests that failed due to page template rendering stuff, now pass again. But we really, really, really should not render PTs in unit tests.16:08
*** tvon has quit IRC16:24
th1aalga:  How was your vacation?16:25
mgedmin|gprsmeeting time?16:30
th1aIndeed.16:30
th1aIs everyone here?16:31
algaAnyone interested in my vacation photos?16:31
algaThat's all I've got to report this time ;-)16:31
th1aGood to have you back, alga.16:31
bskahangood morning16:31
srichterI am here16:31
th1aOK.  Let's get started.16:32
th1aFirst off, I've been working on the new website, which I'm quite happy with (Plone Software Center in particular).16:32
th1aWe've got bskahan working on the site and on the UI contract at the same time however.16:33
th1abskahan:  What's your time frame for skinning the new site?16:33
th1aActually, I guess you should bring everyone up to date on what's done and what needs to be done on that.16:34
bskahansvn, mail and ftp are up, the DNS for staging.st.org, demo.st.org, and ftp.st.org should roll today/tomorrow16:35
bskahanafter the site structure and design are up at the end of the week I'll ask Gus to roll ww.st.org after we're all agreed16:35
th1aOK, so we're shooting for having the new site done and live this time next week?16:36
bskahanmail is just up for outgoing mail at this point since we're not moving the lists16:36
th1aThat's fine.16:36
bskahanyes, ideally next monday everyone can take a look and report bugs and we'll go live shortly after next mondays meeting16:37
* mgedmin|gprs resists the temptation to insert an apostrophe16:37
th1aOK.16:37
th1aI think that we'll do the next round of contracts as sets of proposals within the site.16:38
bskahanI'm a little behind because my dogs decided to eat rat poison over the weekend, otherwise we might have something to lookd at today16:38
*** yvesC has joined #schooltool16:38
th1aEek.16:38
th1aIs everyone ok?16:38
th1aWinston?16:39
bskahanyeah, it sucked16:39
bskahanthey're ok now16:39
bskahannothing like emergency vet trips16:39
th1aThat's good.16:39
th1aAny questions about the web site?16:40
bskahanno, I'll send the sitemap out for review (to the schooltool list) wed/thursday to get feedback16:40
th1amgedmin:  What did you guys accomplish at POV last week?16:40
th1abskahan:  Have you checked out what I've been doing on the site?16:41
mgedmin|gprsoptimisation: great progress (I think), all of our benchmarks (e.g. a view with 5000 events) render in under 1.5 seconds on my (fast) laptop16:41
mgedmin|gprsthere's one more task remaining to clean up and commit16:41
mgedmin|gprsthen we're merging everything into the trunk and release branches16:42
mgedmin|gprssrichter's branch: we switched it with the trunk16:42
mgedmin|gprsgot the tests running (for a while ;)16:42
mgedmin|gprsotherwise didn't take any deep looks inside it yet16:42
mgedmin|gprsignas also found a few bugs16:42
mgedmin|gprsthat sums it up16:42
mgedmin|gprsignas is sick today16:43
th1ajinty: ayt?16:43
th1aSo what's planned for this week?16:43
mgedmin|gprsfinish optimisation16:43
mgedmin|gprsread the checkin messages of all srichter's refactorings16:44
mgedmin|gprslook at the new trunk16:44
jintyth1a: yep16:44
mgedmin|gprsmerge optimisation to trunk16:44
mgedmin|gprsmerge optimisation to release branch16:44
povbot/svn/commits: * jinty committed revision 5014:16:44
povbot/svn/commits: Fix the url of the testbrowser.16:44
algaalso, we had an idea to do a bug day on Friday16:44
mgedmin|gprsstart the next story (sample data) if there's time left16:44
th1aSo... my understanding is that jinty would like to do a release with the optimizations ASAP.16:45
th1aRight jinty?16:45
jintyindeed, there is a possibility that this could then go into breezy16:46
jintybut only a possibility16:46
bskahanthat would be good16:46
* mgedmin|gprs increases the priority of the 'merge into release branch' task16:46
th1aOK, good.16:47
th1aYou could update ("complete work") this when you're done: http://69.60.114.114:8080/schooltool.org-200509/products/schooltool-2006/roadmap/416:49
th1aOK.  srichter.16:49
th1aWhat's up with the UI transition?16:50
srichterEarly last week I worked a bit on the devmode code, making it more stable, providing some initial tests16:50
srichterI also checked in some of the initial REST API documentation attempts16:51
srichterthen, mid-week I started the UI refactoring work16:51
th1aPOV will need some explanation of devmode at soon, since they'll need to use it for the sample data generation.16:51
srichterwe basically put together a proposal first that outlines the choice of software, style guide, etc16:51
srichterthen I started looking at zope.app.pagelet16:52
srichterthe ideas in the pacakge were great, but the implementation was very convoluted, since it made heavy use of macros16:52
bskahanand I've started restructuring existing CSS and documenting it as I go16:52
th1aHave we resolved the capitalization debate?16:53
srichterI since refactored the entire package and it became zope.app.viewlet16:53
bskahanyes16:53
bskahansrichter, will that make it into zope 3.2?16:53
srichterthe refactoring has been approved of by Roger and Philipp16:53
th1aAnd we decided against using all caps in some cases?16:53
srichterwhich gives us a lot of support16:53
srichterbskahan: I don't know. We need the pope's approval16:54
bskahanth1a, I was relying on a browser support bug that marius pointed out16:54
srichterI think that we have a chance16:54
bskahanI still don't concede the readablilty ;)16:54
algait's a matter of preference, IMO16:55
th1aOK.16:55
srichterbtw, <over>16:55
srichteroh yeah, I have also demonstrated how viewlets can be used in the dev mode16:56
srichteri.e. how we do not need to duplicate the view_macros template anymore jsut to insert the devmode16:56
th1aCould the rest of us have a quick summary of what viewlets do for us?16:57
srichterviewlets allow you define regions in a page that are then filled with viewlets16:57
srichtersince the regions retrieve the viewlets from the CA (by default), the construction of the content is decoupled16:58
srichterfor a more detailed set of instructions, read zope.app.viewlet/README.txt16:58
srichterI have put a lot of effort into the documentation there to be simple yet comprehensive of the features and possibilities16:59
th1aAny questions from POV about viewlets at this point?16:59
mgedmin|gprsurl to that README.txt?16:59
mgedmin|gprs(for us lazy ones)16:59
srichterits in the checkout ;-)17:00
mgedmin|gprswhich I do not think I have17:00
mgedmin|gprsand I'm somewhat reluctant to svn up Zope 3 via gprs17:00
mgedmin|gprssrichter, btw buildbot claims that devmode.txt ftest fails in the latest trunk17:00
srichterhttp://svn.zope.org/Zope3/trunk/src/zope/app/viewlet/README.txt?rev=38512&view=auto17:00
mgedmin|gprssimple bug, you need to print browser.contents17:01
mgedmin|gprsnot just >>> browser.contents17:01
srichtermgedmin|gprs: yeah, it does not seem to activate the devmode for ftesting; I have to look at this in a moment17:01
mgedmin|gprssrichter, do you usually run tests before checking stuff in?17:01
mgedmin|gprsoh, my mistake17:01
srichterof course17:01
mgedmin|gprsI thought quoting was the problem17:01
* mgedmin|gprs will be quiet now17:02
th1amgedmin is crabby today.17:02
th1aMoving on...17:02
th1aI've given a lot of thought to how and whether or not to push to keep doing development in branches.17:03
th1aI think it is a good way to work,17:03
th1abut it doesn't seem to fit our structure at this point.17:03
* mgedmin|gprs growls quietly somewhere in the distance17:04
mgedmin|gprsth1a, I will give you my opinion once we merge the optimisation branch17:04
algatrue17:04
th1aOK.17:04
algathe big merges are still ahead17:05
th1aIf getting code reviewed is going to be like pulling teeth, I'd rather spend my energy on other things.17:05
srichterthe merges will be painful :-)17:05
th1aEveryone's going to be diverging in different directions soon anyhow.17:06
th1aPOV doing attendance and demographics,17:06
th1abskahan doing the gradebook,17:07
th1asrichter doing enrollment and the school year workflow.17:07
bskahanthose things are going to be developed as external plugins, not as core features, correct?17:07
th1aYes.  That's what all this refactoring is for.17:08
th1aIn fact, I'm going to set up gradebook, attendance and demographics as separate "Products" on the web site.17:08
th1aSo we can keep the proposals and plans organized.17:09
th1asrichter:  What's next for you?17:10
*** bskahan has quit IRC17:11
srichterth1a: I have to finish the viewlet work a little bit and finish the schooltool demo on it; I am experimenting with a couple different approaches to viewlets17:11
*** bskahan has joined #schooltool17:11
srichterth1a: then I pretty much need new stuff to work on17:11
th1aWell, will that finish 1 and 2 on your proposal?17:13
srichterth1a: yes17:13
th1aAnd you'll be waiting for bskahan to finish the CSS?17:13
srichterth1a: I am stuck with the rest of the proposal to help on until bskahan defines the regions of schooltool17:14
srichterI would prefer if he did the Region definition stuff before the CSS17:14
th1aAh.17:14
srichterhe only has to identify them for me17:14
srichterI'll do the implementation and documentation17:14
bskahansrichter, ok17:14
bskahanI'll do that first17:15
bskahanI should be able to do it tonight/tomorrow17:15
srichterok, cool17:15
th1aI can keep adding content to the web site as is, so the site's not the top priority.17:15
srichterif you want, you can send me screenshots and highlight the regions there17:15
th1abskahan:  will this mostly be based on the proposal for regions that I wrote?17:16
bskahanyes, making wireframes for that17:16
bskahanand identifying the existing (non-poluggable) content regions17:17
srichteryep17:17
th1aOK, so we should be able to keep srichter unblocked.17:17
bskahanyes, I didn't realize he was looking for it until the end of the week, it can be done sooner17:18
th1aOK.  That seems to cover everything.17:19
th1aAnyone have any remaining questions?17:19
*** drwoo has joined #schooltool17:20
th1aConcerns?17:20
algabug day17:20
jintyth1a: comments on your roadmap17:20
th1aalga:  When?17:20
algaif we push the issue tracker on Friday, will people be available to help resolving issues that "belong" to them?17:20
bskahanwould it be possible to move the monday meeting time ahead an hour?17:20
algabskahan: that would be bad for me17:21
bskahanok17:21
algaI have a lecture starting at 15:00 UTC17:21
th1abskahan:  that would be bad for me, too.  I'm barely awake as it is.17:21
bskahanlol17:21
bskahannm then17:21
algawait a minute17:21
algawhat do you mean by 'ahead'?17:21
th1aearlier?17:22
bskahanearlier17:22
algaearlier is fine for me17:22
bskahanbut th1a isn't awake ;)17:22
th1aThat would be rough for me.17:22
algaok, i very much respect that :-)17:23
bskahannot a big deal then17:23
th1aI had an easier time getting up in Vilnius.17:23
th1aGetting back to alga's question: do you want to have a bug day on Friday?17:24
th1aJennifer was lamenting the inferiority of American mushrooms (to Lithuanian ones) yesterday.17:25
th1aI put some capers on our pizza.  It was just like being in Vilnius.17:25
bskahanI think its a good idea, I'm not sure about my availabilty but I'm willing to be assigned bugs while absent to tackle over the weekends17:25
algaWe (POV) are getting paid for herding bugs, so we will be doing some bug hunting anyway17:25
algabskahan: OK17:26
algath1a: the Lithuanian sea side I recommended you to visit: http://www.akl.lt/alga/photos/Family/atostogos17:26
th1aIs that where you were last week?17:27
algayep17:27
th1aNice.17:27
algaone British gentleman said it felt a lot like the coast at the Scottish Highlands17:27
algaminus the rocks17:27
th1aOK.  I think we're done here.17:27
th1aOr did I forget jinty.17:28
th1ajinty:  What was your question?17:28
th1aWell, I'll end the "official" meeting on time.17:29
* th1a bangs the virtual gavel.17:29
jintyth1a: a risk factor in you roadmap17:29
th1aYes?17:29
jintyif Zope3.2 is not released on time, I will not be able to build packages17:29
jintyif that makes a difference17:30
jintythis is due to the new way the packages are built. i.e. they depend on the zope3 package rather than include zope317:30
th1ajinty:  OK.  Not much we can do about that.17:31
jintyjust letting you know17:31
th1ajinty:  Thanks.  We'll just have to twist the arm of the Zope3 release manager to get it out on time.17:33
* mgedmin|gprs looks at the new plone site and wonders if he's an account there17:33
th1aYou'll need to create new accounts :-(17:34
jintyotherwise, I am uncomfortable with schooltool depending on parts of Zope3 that are not included in Zope3 tarballs.17:34
jintyanyway, I must be off to shower and a birthday party;)17:34
jinty;)17:34
th1aMigrating them didn't work automatically.17:34
mgedmin|gprsjinty, have fun17:34
th1aSee you, jinty.17:34
jintyadios17:35
*** jinty has quit IRC17:35
th1abskahan:  I'll need to set up a SchoolTool instance for FHS this week, if possible.17:36
th1aHave you worked out the chroot problem?17:36
*** alga has quit IRC17:42
bskahanth1a, what OS is the server running there?17:47
th1aNo, I mean, I need to set up an instance for them on our server.17:48
th1aThey don't have a server.  That's the problem.17:48
bskahanah, didn't understand that17:49
bskahanno, having debian debootstrap issues on our server.  someone in ubuntu suggested using rootstrap instead so I'm going to try that17:50
bskahanwe need more ram17:50
*** drwoo has quit IRC17:50
th1aOK.  1 gig?17:51
bskahan217:51
th1a2 gigs total?17:51
bskahanwe're going to have plone and minimum 2 schooltool instances running17:51
bskahanyes17:51
bskahan1.5 more than we have17:52
th1aWhat are the memory requirements of SchoolTool, then?  512 mb?17:52
*** mgedmin|gprs has quit IRC17:52
bskahanI'll set up your chroot when I set up the schooltool demo chroot17:53
bskahanthat's a good question17:53
bskahanbut I would say 512 is a starting point17:53
th1aOK.  Thanks.  I'll request more memory.17:54
bskahana good goal (in keeping with the initial project goals) is probably to keep the memory requirements for a school of 1k students to 512M of ram tops, and shoot for 25617:59
*** yvesC has quit IRC18:00
bskahanafter the sample data story is done we could probably set up some good benchmarks with apache bench18:00
th1abskahan:  We probably won't do memory optimization until next summer.18:01
bskahansince "What hardware do I need" is going to become the #1 question18:01
th1aYes, answering that question accurately is more important initially than keeping the requirements low.18:02
* bskahan nods18:02
*** mgedmin has joined #schooltool18:06
*** Aiste has joined #schooltool18:08
bskahansiege -u desktop:7080 -d1 -r4 -c2518:15
bskahanon an empty instance is pretty fast, but it logs quiet a few zodb errors18:16
th1asiege?18:16
bskahanwell, not quiet a few18:16
bskahanbut a few18:17
th1aquite?18:17
bskahanthanks18:17
bskahanwho needs to look at what they're typOing18:17
th1aWhat is siege?18:18
bskahanweb page benchmarking tool18:18
th1aAh.18:18
bskahanthat's 25 concurrent users hitting once a second, 4 times18:19
bskahanit's a trivial test though, since it's just an empty school calendar18:21
*** bskahan has quit IRC18:23
mgedminthose are ZODB Conflict Errors I assume18:26
povbot/svn/commits: * srichter committed revision 5015:18:57
povbot/svn/commits: Switched to a different method of developing often-recurring viewlets that have only small differences. Now javascript and CSS links are inserted into a page using special viewlets that make it very simple to do the configurstion.18:57
*** maharaja has joined #schooltool18:58
mgedminsrichter, did you fix functional tests?18:59
mgedminyay, you did!19:28
*** maharaja has quit IRC19:43
*** jinty has joined #schooltool20:13
povbot/svn/commits: * mg committed revision 5016:20:23
povbot/svn/commits: Measure CPU time, not wall time.20:23
*** thisfred has quit IRC20:56
povbot/svn/commits: * mg committed revision 5017:21:20
povbot/svn/commits: Added "smart" caching for CalendarViewBase.getDays.  ~10% performance gain for nonrecurrent event benchmark.  A view knows which time range is of interest, so it can call calendar.expand only once for this range, instead of calling calendar.expand five times for smaller, overlapping time ranges.21:20
povbot/svn/commits: I think that local caches in getMonth/getYear, and perhaps in dayEvents also, are now not necessary and could be removed.21:20
povbot/svn/commits: Added an XXX about a bug (issue 373 -- calendar boundaries are in UTC).21:20
povbot/svn/commits: Note that the many-recursive-events benchmark shows a slow-down of the yearly view (other views show speedups).  This is very very strange.  I have discovered that with that setup, calling calendar.expand for the time range from 2004-11-29 to 2006-01-02 gives me some 118000 events.  However, calling calendar.expand for several smaller overlapping time ranges (the union of21:20
povbot/svn/commits: which is exactly the same 2004-11-29..2006-01-02 range) gives me LESS events in total (some 113000).  How can this be?  Is there a bug somewhere?21:20
povbot/svn/commits: Note that the four recursive events benchmark shows a slowdown, but I consider that acceptable -- although I would like to determine the cause.  It could be the same problem that causes a slowdown with many recurrent events in the yearly view.21:20
mgedminAAAAAAAAAARGH!21:23
mgedminignas, I will personally throttle you!!!21:24
mgedminI've been debugging the mysterious benchmark slowdown for an hour21:24
mgedminand only now I notice21:24
mgedminthat you carefully create a random number generator with a fixed seed for reproductivity21:24
mgedminand then just use random.choice/random.randrange!21:24
mgedmincompletely ignoring that generator21:25
* mgedmin now knows why benchmarks were so hard to reproduce21:25
Aiste:))21:27
povbot/svn/commits: * mg committed revision 5018:21:28
povbot/svn/commits: Make benchmarks deterministic, not randomized!!!21:28
* jinty ply's mgedmin's hands from ignas' throat21:34
*** bskahan has joined #schooltool21:36
* mgedmin puts the voodoo doll of ignas away and prepares to drive home21:45
ignaswhat ? where am i ?21:46
*** mgedmin has quit IRC21:46
*** bskahan has quit IRC22:24
*** bskahan has joined #schooltool22:43
povbot/svn/commits: * jinty committed revision 5019:22:50
povbot/svn/commits: A recent change to zpkgtools breaks SchoolTool-test.py with this error:22:50
povbot/svn/commits: tar: SchoolTool-0.0.0/Dependencies/zope.cachedescriptors-SchoolTool-0.0.0/setup.cfg: Cannot change ownership to uid 0, gid 0: Operation not permitted22:51
povbot/svn/commits: My guess is that it tries to create files with root ownership, even though 'info tar' says that the '-o' option only has an effect when used by the superuser. HURMFFF!22:51
povbot/svn/commits: Fortunately removing this 'o' makes the error go away, exposing other errors.22:51
jintysrichter: am I right that I probably need to add zope.app.viewlet to src/schoolbell/app/DEPENDENCIES.cfg?22:52
*** bskahan has quit IRC23:06
*** tvon has joined #schooltool23:09

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