IRC log of #zope3-dev for Thursday, 2006-04-27

*** zbir has quit IRC00:03
*** sashav has quit IRC00:04
*** sashav has joined #zope3-dev00:04
*** whit_ has joined #zope3-dev00:07
*** whit_ has quit IRC00:09
*** whit_ has joined #zope3-dev00:14
*** tarek has quit IRC00:15
*** jinty has quit IRC00:18
*** ChrisW has joined #zope3-dev00:19
ChrisWj1m: ping?00:21
*** natea has joined #zope3-dev00:21
ChrisW(or anyone else with zodb foo, for that matter)00:21
ChrisWI'm trying to work on
ChrisW(not zope 3, I know, so please point me to a better channel if there is one...)00:21
*** benji has quit IRC00:22
ChrisWgot down to around line 149 in, which is roughly
ChrisWthe base.__setstate__(state) works00:23
ChrisWbut some tiem after that the changes get lost :-S00:23
ChrisWthis appears to be wondering into deep ZODB fu, was wondering if you had any pointers00:23
ChrisWand, more trickilly, how to write tests for this...00:23
j1mI don't have time to help with this.00:23
ChrisWno worries, any pointers (including to people) greatfully received00:24
*** whit_ has quit IRC00:26
*** whit has joined #zope3-dev00:27
*** sashav has quit IRC00:30
*** sashav has joined #zope3-dev00:30
*** dobee has quit IRC00:37
*** natea has quit IRC00:41
*** zbir` is now known as zbir00:42
*** natea has joined #zope3-dev00:43
*** efrerich has quit IRC00:47
*** sashav has quit IRC00:55
*** s1 has joined #zope3-dev01:19
*** zbir has quit IRC01:19
*** s1 is now known as sm01:19
*** whit has quit IRC01:22
*** jinty has joined #zope3-dev01:23
*** whit has joined #zope3-dev01:28
*** ChanServ sets mode: +o tav01:30
*** whit has quit IRC01:32
*** natea has quit IRC01:36
*** ChrisW has left #zope3-dev01:47
*** natea has joined #zope3-dev01:49
*** SiggyF has quit IRC02:00
*** sm has quit IRC02:01
*** jinty has quit IRC02:04
*** jhklm4 has joined #zope3-dev02:08
*** sm has joined #zope3-dev02:09
*** whit has joined #zope3-dev02:25
*** rockyburt is now known as rockyburt|away02:29
*** whit has quit IRC02:29
*** zbir has joined #zope3-dev02:31
*** jhklm4 has quit IRC02:45
*** natea has quit IRC03:00
*** roym has quit IRC03:30
*** Londo_ has quit IRC03:43
*** rockyburt|away has quit IRC03:44
*** seang has quit IRC03:46
*** natea has joined #zope3-dev04:15
*** natea has quit IRC04:34
*** stub has joined #zope3-dev04:39
*** RaFromBRC has quit IRC05:11
*** niemeyer has quit IRC06:25
*** tristil has joined #zope3-dev06:45
*** dobee has joined #zope3-dev07:12
*** stub has quit IRC07:22
*** sashav has joined #zope3-dev07:56
*** tristil has quit IRC08:04
*** sashav has quit IRC08:12
*** stub has joined #zope3-dev08:13
*** eins has joined #zope3-dev08:14
*** mexiKON has joined #zope3-dev08:41
*** zopePloneConsult has joined #zope3-dev08:41
*** philiKON has quit IRC08:49
*** stub has quit IRC08:53
*** hdima has joined #zope3-dev08:59
*** sm has quit IRC09:13
*** d2m has quit IRC09:17
*** d2m has joined #zope3-dev09:30
*** zagy has joined #zope3-dev09:41
*** stub has joined #zope3-dev09:43
*** povbot has joined #zope3-dev14:49
srichterno, that's the way to do it14:49
*** natea has joined #zope3-dev14:51
regebroHmmm. I guess I should do this in my default IBrowserPublisher.browserDefault method.... Yeah, that should work.14:53
srichtermmh, SelectedManagementView.html does not seem to work correctly anymore :-(14:56
faassenhm, NotFoundError is gone?14:59
faassenor is that one back now?14:59
faassenI wonder where NotFoundError went.15:00
*** andres_ has quit IRC15:00
faassenhm, there is a deprecation warning in Zope 3.215:01
faassenbut somehow that never fired.15:01
faassenit says I should use standard exceptions instead, but which one?15:02
regebrofaassen: TraversalError, no?15:02
srichterfaassen: Jim answered this question above15:02
srichterhe claimed it should be LookupError15:03
srichterthough the notfound views are registered on TraversalError15:03
srichterI had the same problem with my app here15:03
faassenhm, problems in optionstorage.15:03
faassenschooltool apparently uses this.15:04
faassenand this still uses NotFoundError.15:04
faassenit claims the following:15:04
faassenclass StorageNameNotFoundError(NotFoundError):15:04
faassen    # BBB: Can be Remove in 3.315:04
faassen    pass15:04
faassenbut then it uses StorageNameNotFoundError above.15:04
faassenI mean, below.15:04
srichtermmh, unfortunately, the deprecated module did not work too well15:05
faassenShould I replace this with a lookup error or something?15:05
faassenI don't even know what optionstorage is supposed to do. :)15:06
srichteroption storages are basically manageble vocabularies15:06
srichterthey allow you to manage the terms in the ZODB15:06
srichteralso supports i18n15:06
faassenthe gradebook uses them. :)15:07
srichterunfortunately they do not have tests, but it was easier to use them than to rewrite the code and test it15:07
faassenI just wonder what uses StorageNameNotFoundError..15:07
srichterI dunno15:08
srichterI would try to remove it and see what happens15:08
faassenwell, something raises it. :)15:08
faassenI'd need to raise something instead.15:08
faassennothing appears to be using it.15:09
faassenbut I'd need to replace that exception with some other.15:09
srichterI think a LookupError will be fine15:09
faassenok, I shall try that.15:09
srichterso just inherit from LookupError15:09
faassenhm, stuff is raising LookupError but it's never imported.15:10
faassenwhat's up with that?15:10
faassenoh, python builtin.15:10
efgeyeah little know builtin15:10
faassenokay, this is a bit of wild west coding, just taking out the exception altogether. :)15:11
*** b_52Centos is now known as b_52Centos_miam15:19
*** ignas_ has quit IRC15:19
*** ignas has joined #zope3-dev15:19
*** MacYET has joined #zope3-dev15:20
*** ignas has quit IRC15:20
*** ignas has joined #zope3-dev15:20
*** mgedmin_ has joined #zope3-dev15:21
*** mgedmin_ has quit IRC15:22
faassenhm, AdapterRegistration doesn't have a value attribute anymore.15:23
srichterthat's a bug it seems15:23
srichtermaybe it has been replaced by another attribute15:23
srichterit is called factory now15:24
faassenthat attribute?15:24
srichter"value" used to be the adapter factory (as far as I remember); the interface now says it is called "factory"15:24
faassenso what do we do? add an attribute value and put in a deprecation warning?15:25
faassenhey MacYET15:25
srichterfaassen: yes, that's what I would do15:25
MacYETsrichter: you're the release slave for 3.3?15:25
faassenthere is apparently also a component attribute.15:25
srichterMacYET: probably so ;-)15:25
faassenthat got remapped to factory.15:25
faassensrichter: so how does that work?15:25
faassensrichter: I mean, if component and value were the same thing before, what I would do is right.15:25
*** jukart has joined #zope3-dev15:26
jukartdobee, ayt?15:27
srichterfaassen: all you really need is a property called "value" that raises a deprecation warning when accessed15:27
faassenI know, but since there is already an attribute component.15:27
faassenthat does just that.15:27
srichterI think zope.depcrecation has a helper function15:27
faassenI'm trying to figure out whether I'm doing the right thing.15:27
faassenyeah, I can copy and paste literally from 'component'15:27
faassenbut I am trying to figure out where that attribute came from.15:27
faassenit didn'te xist on AdapterRegistration in zope 3.215:28
faassenso I wonder whether that was a mistake.15:28
srichterit should have15:28
faassenI suspect so, as there's a component attribute on UtilityRegistration.15:28
srichterMacYET: do you have suggestions for the release schedule15:28
faassenin 3.215:28
srichterit is still there in the trunk15:29
dobeejukart: jep15:29
MacYETbasically depends on how long lennart needs for testing15:29
MacYETregebro: ping15:29
faassensrichter: I suspect that this deprecation warning is a bug, the one for component.15:29
faassensrichter: but since it's not doing any harm I'll leave it in for now.15:29
regebroMacYET: Hi!15:30
MacYETtalking of the evil :)15:30
faassenevil, huh?15:30
MacYETwhat is our strategy for merging the two branches?15:30
regebroWell... 1. Deprecate five:traversable and five:defaultViewable (I'm doing that now).15:31
regebro2. Test with CPS.15:31
faassenhm, doctest signature of AdapterRegistration also changed.15:32
regebro3. Merge. :)15:32
MacYETcan you check it with plohn as well?15:32
srichterfaassen: yeah, those type of changes we just have to live with15:32
faassensrichter: ok..15:32
regebroMacYET: I'll try.15:33
regebroI have Plone 2.5 checked out, I assume thats the version to test with.15:33
MacYETi know, a pita for CPS guys :)15:33
regebroIf it works with CPS and not with Plone, I'll see that as a Plone bug. :P15:34
MacYETplohne is always the guilty15:34
regebroAnyway, Plone 2.5s recommended Zope version is still 2.8, so it will be some time before they get into the 2.10 train.15:34
MacYETsricher: for the release: either 2.or3. or after 9th ?15:34
MacYETyou'll never know which train they take15:35
srichterMacYET: I think beta 1 on May 2?15:35
MacYETno problem for me15:35
MacYETregebro: fine with u?15:35
srichterMacYET: actually, I am in transit on that day, so for me it would have to be May 315:35
MacYET2-3 is fine15:35
MacYETi'll be on the linuxtag on 4th...15:36
srichterMacYET: then beta 2 after the 9th sounds good15:36
srichterI think we should aim for RC 1 around the 20th?15:36
MacYETand final 1.615:37
regebroSounds good.15:38
*** niemeyer has joined #zope3-dev15:38
MacYETsrichter: shall i post the schedule?15:39
srichteryeah, that would be great!15:39
MacYETwill do15:39
*** _sussi has joined #zope3-dev15:44
*** andres has joined #zope3-dev16:00
*** alecm has joined #zope3-dev16:01
*** zbir has quit IRC16:03
*** rockyburt has quit IRC16:03
*** stub has quit IRC16:04
*** b_52Centos_miam is now known as b_52Centos16:09
*** MJ has joined #zope3-dev16:10
*** zbir has joined #zope3-dev16:12
*** tav has joined #zope3-dev16:13
*** rockyburt has joined #zope3-dev16:21
*** zbir has quit IRC16:22
efgebeta? has the schedule moved a month earlier? I thought we had feature freeze may 1st, beta june 1st, and release july 1st.16:24
efgeyou guys want to do a beta at the same time as the feature freeze?16:25
j1mhey dudes16:26
faassenhey j1m16:26
j1mefge, that is theoretically the idea of the feature freeze.16:26
j1manything checked into the trunk should be ready for beta, or at least, the author should think it is.16:27
efgei see16:27
j1mGiven how late I merged my branch, I think we may need to slip the freeze.16:27
faassenwhat features will get in later then?16:28
j1mI dunno, but I had the ipression that other work was blocked by the merge.16:28
faassenpossibly Five work, I haven't kept track of that.16:28
j1mfaassen, right, I'm assuming mainly Five work16:28
j1mI don't have anything, so it's up to others to speak up.16:28
j1mBut I'm aware that the late date of the merge may have put others in a bad spot.16:29
j1mmexiKON, ayt?16:29
mexiKONjust got home16:29
*** benji has joined #zope3-dev16:29
mexiKONwhat's up?16:29
*** BjornT has quit IRC16:30
srichterj1m: I committed a test and a fix to the site deletion problem; please have a look to make sure it is correct16:31
faassencould there be any reason why setHeader on response somehow changed after the merge?16:31
j1mmexiKON, efge are you aware of any Five issues with a May 1 feature freez?16:31
j1msrichter, k16:31
mexiKONthere's the philikon-local-components branch of Zope 2 and Five16:31
mexiKONthey deal with the jim-adapter branch16:31
mexiKONthey need to be merged if zope 2.10 wants to use zope 3.316:31
faassenbecause at first glance it looks like setHeader is not doing anything.16:32
mexiKONthey're not 100% done yet16:32
mexiKONi won't have time to fix them before may 1st16:32
efgej1m: yes my issues were w.r.t the branches that will make up zope 2.10, which have to be tested with zope 3.316:33
mexiKONright. there's the ajung-zpt-integration branch and regebro's branches16:34
mexiKONand my philikon-local-components branch16:34
*** MJ has quit IRC16:34
j1mso, we (royal we ;) need to sort that out before we can finalize the release schedule.16:36
faassenwhat do regebro's branches do?16:37
j1mIMO, if y'all have something important that you are sure you could finish if you had an extra week, I'd be OK with slipping a week.16:37
*** zbir has joined #zope3-dev16:37
j1mIf not, then we just won't have it (whatever it is) in this release.16:38
mexiKONfaassen, see his psoting to zope-dev16:38
MacYETfaasen: refactoring of the traversal stuff + zpublisher/twisted integration16:38
mexiKONj1m, 1 extra week is all i shall need16:38
regebrofaassen: Using IPublishTRaversable during Zope2 traversal + hopefully browserDefault stuff.16:39
regebroand put a WSGI interface in ZPublisher.16:39
faassentwisted integration in Zope 2 will land in Zope 2.10?16:39
faassenpresumably not as the default. :)16:40
regebroI'm not gonna have time to do that.16:40
j1mMy main input is to point out that the special circumstances of the late merge might justify a freeze delay.  But if we agree to the delay, we need to be committed to making up the time.  Out target should still be an early June release.16:40
regebroI though deprecating the five:defaultViewable statement would be easy....16:40
mexiKONregebro, it doesn't have to be all in this reelease16:41
MacYETmexiKON: define all16:41
mexiKONall of the traversal refactoring... having an alternative to __bobo_traverse__ in form of IPublishTraverse adapters is quite an accomplishment already16:42
MacYETthen the rest for 2.1116:43
mexiKONi would gues sso16:43
*** alecm has quit IRC16:43
j1mbe conservative folks16:44
*** jinty has joined #zope3-dev16:44
j1mdon't overpromise16:44
MacYETespecially don't touch the build process this time :)16:44
j1mIf you aren't *sure* that you can get something in in a releaseable state, then don't try.16:44
j1mMacYET, I was hoping that we could get zpkg out of the build process this time. :)  But I don't think we got far enough.16:45
MacYETdidn't we introdruce zpkg in 2.9?16:45
faassenj1m: I wish we got an alternative to acquisition for Zope 2 security checks, location based security would've been nice.16:46
j1mmexiKON, how much stuff remains in that is needed by Five?16:46
j1mfaassen, yup16:46
mexiKONj1m, i haven't done the metrics yet. i rahter spent time on fixing Five to work with the jim-adapter branch in the first place16:46
faassenj1m: plenty if is still in use, which it is.16:46
mexiKONj1m, i'll post some statistics after the freeze16:46
mexiKONright16:47 wasn't the target of PackageGeddon2. as it was faassen's playground anyways (re: zope.widget ;))16:47
j1mmexiKON, it was an error on my part to do the shrunkage work on the jim-adapter branch. Oh well.16:47
*** tav has quit IRC16:47
mexiKONi don't think there was a right or wrong solution :) both ways would have been problematic, one way or the other16:48
*** stainsby has joined #zope3-dev16:48
mexiKONanyways, it's finally in the trunk now16:48
j1mexcept that I noticed that there wasn't as much overlap as I expected.16:48
j1mIs there a Zope 2.10 wiki page?16:49
faassenmexiKON: zope.widget I'm not going to tackle, as Gary said they had all kinds of advanced plans for a rewrite.16:49
faassenmexiKON: so it won't happen.16:49
mexiKONi'm not touching :)16:49
j1mwhat's the big deal with
j1mshould be easy to just move it outta defines tons of widgetgs.16:50
j1mI would think.16:50
faassenyes, but I proposed doing it.16:50
faassenat least moving out the widgets.16:50
faasseninto zope.widget16:50
MacYETj1m: yes, but it's outdated16:50
faassenbut then Gary said we have plans for zope.widget, rewrite complete, etc.16:50
mexiKONMacYET, not entirely16:50
faassenso then I stopped with that.16:50
mexiKONMacYET, the widgets we still need16:50
MacYETwidgets?16:50 defines all the widgets we currently have in z316:51
j1mI'd rather like to help get zope 2 to *not* include
mexiKONj1m, we won't get it all done in this release16:51
j1mI'd be willing to tackle, if only by moving it to zope.form.16:51
mexiKONj1m, let's try to fix everything up so that we can make the release and worry about this for zope 2.1116:51 won't hurt much16:51
faassenI don't think it's realistic to tackle all the move this release.16:51
mexiKONj1m, as you said, let's be conservative :)16:52
faassenyeah. :)16:52
j1mI just didn't know what else was left in that z2 needed.16:52
* faassen does the Use Your Own Words Against You move with mexiKON16:52
mexiKONfaassen, ahem :)16:52
faassenhey, you were in China for years to learn kung fu, right?16:52
mexiKONi meant this strictly release-process-oriented :)16:52
j1ma straight move of packages wo persistent objects should be easy, as long as we don't try to refactor at the same time.16:53
mexiKONfaassen, yes, i'm a triple master in gong fu16:53
mexiKONfaassen, haya!16:53
* faassen is deadified by mexiKON's skillz.16:53
mexiKONj1m, the devil is in the details16:53
faassenanyway, yeah, that's true j1m16:53
mexiKONj1m, ZPublisher and five import some things from
faassenbut I think we should just not worry too much about making five not rely on yet this round.16:53
mexiKONit's like pulling lots of strings16:53
j1mmexiKON, it would good to get a rough idea of what else beside is needed.16:54
mexiKONif it's that urgent, i can compile a list later16:54
j1mI'm confident that I could knock off this weekend.16:54
mexiKONwhat about only moving the widgets out to zope.widget?16:54
j1mdo you have any intuition for how long the list is?16:54
mexiKONthe rest of can die for all i care16:54
* regebro is a quadruple master of fudge ju.16:55
j1mmexiKON, that would be harder and riskier.16:55
j1mfwiw, for, I would:16:55
j1m- Move it to zope.form16:55
j1m- Leave an alias *without* deprecation warnings.16:55
mexiKONwhat good would that do?16:56
faassenj1m: you'd still keep the widgets in, right?16:56
j1mThis way, we would not make people change their code until we are ready to have them change it to something more permanent16:56
j1mBut then five would have one big thing that is no longer in
mexiKONwell, five does use the widgets...16:57
faassenand it also uses other bits and pieces I think.16:57
faassenfor addform and editform.16:57
* mexiKON compiles a list16:57
j1mIf there wasn't much left in that was needed, then we could stop using in z3.16:57
j1massuming that we could move the other things out easily.16:57
mexiKONbiggest piece left in five is probably also because of ISIte16:57
* MacYET heads for the airport, bye16:58 because of view metaconfiguration stuff16:58
*** MacYET has left #zope3-dev16:58
mexiKONthe latter is actually in my proposal16:58
mexiKONi just hadn't had the tiem to move it16:58
mexiKONit also uses for placelesssetup, but i'm convinced that 95% of the tests that use it don't really need it16:59
j1mOK, so, something for y'all to think about.  My offer stands to do
mexiKONi appreciate the offer, but it wouldn't get us *that* much closer to knocking in z216:59
j1mthen let's drop it.17:00
*** zopePloneConsult has left #zope3-dev17:05
*** Aiste has quit IRC17:08
mexiKONfaassen, saw my reply regarding IVocabularyFactory?17:08
mexiKONbbl. jogging17:10
faassenmexiKON: oh, no, I will check.17:11
faassenI realized that it didn't exist in Zope 3.2.17:11
mexiKONi'll be back in 1 hour17:11
mexiKONor so17:11
faassenafter that.17:11
faassenhowever, schooltool was using it so I figured it couldn't hurt. :)17:11
faassenit's a pretty harmless warning.17:12
*** eins has quit IRC17:14
*** hdima has quit IRC17:15
*** jukart has left #zope3-dev17:17
*** sm has joined #zope3-dev17:18
*** BjornT has joined #zope3-dev17:18
*** stainsby has left #zope3-dev17:20
j1mIt would be good for someone to make a 2.10 release page and try to gather what needs to be done.17:33
j1mIf anyone is going to work on collector entries, they should assign them to themselves first.17:34
j1mso we don't duplicate efforts.17:35
*** vlado has joined #zope3-dev17:36
*** andres_ has joined #zope3-dev17:41
faassenj1m: by the way, I'd prefer to ship with Zope 2.10 no matter whether we can make five not depend on anything in it anymore.17:43
faassenj1m: as we tend to have a lot of synergy of people using new bits of Zope 3 and that slowly migrates into Five.17:44
faassenj1m: and you'd block this.17:44
faassenj1m: in the history of five we've had lots of new Zope 3 features exposed during Zope 2.x releases.17:44
faassenj1m: most recently for instance formlib (though granted this was in zope already, not
faassenI mean, between Zope 2.x rleeases.17:45
faassenjust in normal development on top of Zope 2.8 or Zope 2.9.17:45
*** sashav has quit IRC17:46
j1mThere's no reason someone couldn't still to this in development mode.17:46
j1mBut I'm not going to argue the point.17:46
faassenpeople are often motivated by short-term gains.17:46
j1mIt doesn't effect me.17:46
faasseni.e. do it within the Zope 2.x release cycle.17:46
regebroCPSMailAccess for example uses and't.know.what. :)17:46
faassenthat's another risk, you risk breaking Five-based applications that expose more of Zope 3 than Five does by itself.17:47
srichterI agree that Zope 2 should zhip with all of Zope 317:47
srichteruntil we are sure that really only Zope 3 app server speciifc stuff is in there, which is not much17:48
srichtermaybe the skins, server package, twisted package and appsetup package17:48
*** gumpa has joined #zope3-dev17:50
faassenj1m: we had a thread some time ago on zope-dev I think, where I tried to make the case. :)17:50
*** andres has quit IRC17:58
*** jukart has joined #zope3-dev18:00
regebroI'm trying to provide an adapter for "*" in python (it's a test).18:01
regebroAnybody know how to do that?18:01
jukartmaybe just adapt to Interface18:01
regebroGood idea, lets try.18:02
*** vlado has quit IRC18:03
regebroYeah, that worked, thanks jukart.18:03
*** Aiste has joined #zope3-dev18:15
*** alga has quit IRC18:18
*** Aiste has quit IRC18:20
*** whit has joined #zope3-dev18:40
*** whit has quit IRC18:44
*** srichter has quit IRC18:55
*** zbir` has joined #zope3-dev19:11
*** zbir has quit IRC19:19
*** zagy has quit IRC19:22
regebroHow are you supposed to figure out what the error is, when you have a functional doc test that fails?19:24
regebroAll you GET is internal server error...19:25
mexiKONthat's why testbrowser tests are a pain to debug19:25
regebroI'm grasping blindly...19:26
mexiKONregebro, my way of trouble shooting these things has been so far to add the same HTTP call as a traiditional http(...) call19:27
BjornTregebro: you'll get a traceback instead if you set browser.handleErrors to False19:27
mexiKONthen you'll get the traceback19:27
mexiKONBjornT, ahh19:27
* mexiKON should've read the testbrowser docs :)19:28
mexiKONwhy isn't that on by default?19:28
mexiKONj1m, iv'e compiled a list of remaining tasks for zope 2.10. see my posting to zope-dev.19:29
regebroOh THANK YOU!19:29
regebroWhat a relief. :)19:30
* mexiKON has a coiuple of tests to try that on himself19:30
regebroOh, yeah, right, Five monkeypatches Expression.restrictedtraverse.... That's why I couldn't put a set_trace() there...19:31
*** _sussi has quit IRC19:31
*** _sussi has joined #zope3-dev19:32
mexiKONit does?19:32
mexiKONthat might be the reason why ajung's branch is failing too19:33
mexiKONfaassen, still there?19:33
*** _sussi has quit IRC19:34
faassenmexiKON: yes, here.19:35
mexiKONfaassen, the alias you made for IVocabularyFactory doesn't say when it'll disappear. how about we say that it'll disappear in z 3.3 final19:36
mexiKONshould give schooltool etc. enough time to adjust19:36
faassenmexiKON: ok, sure.19:36
faassenmexiKON: schooltool is done already. :)19:36
faassenmexiKON: so I guess it can be taken out again. :)19:37
* mexiKON takes it out then19:37
mexiKONok done19:38
*** mexiKON is now known as philiKON19:38
*** seang has joined #zope3-dev19:38
*** _sussi has joined #zope3-dev19:39
*** sashav has joined #zope3-dev19:43
*** jinty has left #zope3-dev19:46
*** jinty has quit IRC19:46
j1mHm, I just got a chance to look at the colector and there aren't any outstanding issues.20:00
j1mCool. :)20:00
*** zagy has joined #zope3-dev20:03
*** zagy has quit IRC20:21
*** _sussi has quit IRC20:21
*** RaFromBRC has joined #zope3-dev20:23
*** tarek has joined #zope3-dev20:29
j1mfaassen, where did you get a value attr on adapter registrations?20:40
faassenj1m: there is this package schooltool.devmode that Stephan wrote20:42
*** niemeyer has quit IRC20:42
faassenj1m: and it was accessing that.20:42
j1mThat's weird, since adapter registrations didn't have a value attr before the merge.20:42
faassenclass AdapterRegistration(object):     """Registration for a simple adapter."""      def __init__(self, required, provided, name, value, doc=''):         (self.required, self.provided,, self.value, self.doc          ) = required, provided, name, value, doc20:43
faassenin Zope 3.220:43
faassenself.doc also was gone, replaced with self.info20:43
j1mYeah, good catch20:44
faassenanyway, that code has a self.value20:44
philiKONj1m, i remember that apidoc was using .value, .doc on adapter registrations20:44
faassenthis is apidoc like code.20:44
faassenI think.20:44
faassenanyway, after that the tests ran again. :)20:44
faassenI have no idea what the code actually does.20:44
j1mwhat file did the code above come from?20:44
faassenI just spent most of the day chasing breakages and most of all fixing lots of deprecation warnings in schooltool.20:45
faassenso zope.component.site20:45
faassenanyway, going home now.20:46
faassensee you!20:46
*** niemeyer has joined #zope3-dev20:46
*** faassen has quit IRC20:46
j1mOh yeah20:46
*** admp has joined #zope3-dev20:48
ignasj1m: did someone add caching mechanism to multi adapter queries ?20:53
*** andres_ is now known as andres20:56
philiKONignas, no. the speedup in the adapter lookup is mostly due to a different strategy of looking them up21:00
ignasphiliKON: well - I have just had a wonderful experience in a functional test that was registering a new view through provideAdapter21:01
ignasphiliKON: and when one queries the url - gets one object, then registers a more specific view - gets the same old view one it looks like something is geting cached21:02
ignass/view one/view21:02
philiKONdefine more specific21:02
*** tarek has quit IRC21:02
ignasi'll try21:03
ignasin a functional test i have a ISchoolToolApplication as root21:03
ignasif i do "POST /" register view, do some requests "POST /" - it works as expected21:04
ignassecond POST goes to the new view21:04
ignasbut if i do21:04
ignassome requests, "POST /", register view, "POST /"21:04
ignasthe post goes to the old view21:05
j1mignas, yes, all adapter queries are cached21:05
philiKONj1m, hah. didn't know that21:05
j1mmulti-adapter queries are more cached than they used to be.21:05
philiKONignas, by the way, it is debatable whether functional tests should register new things as they go along21:05
ignasphiliKON: i know21:06
philiKONignas, it's better to register everything in an ftesting.zcml file beforehadn21:06
j1mregistering a new view should cause the cache to be invalidated.21:06
ignasztapi.provideAdapter((ISchoolToolApplication, IHTTPRequest), Interface, BadView, 'POST')21:06
ignasmaybe the registration is wrong in some way?21:06
j1mThis is so far out of context I have no idea21:07
philiKONseems to look ok21:07
philiKONi recommend using zope.component.provideAdapter, by the way21:07
philiKONthe ztapi arrangement of arguments is quite incomprehensible21:07
j1mplus, we'd like to deprecate ztapi at some point.21:08
j1mCan you show both registrations?21:09
ignasthere is only one21:09
j1mThat isn't consistent with what you said above.21:09
ignasthe original view is a normal view registered in the application zcml21:09
j1mYou said you registered a new view.21:10
j1mOK, how was it registered?21:10
ignasthe registation is for POST though21:13
ignasand the first POST was to check for Method Not Alowed error21:13
ignasthe *new registration* i mean21:13
j1mI have no idea what you're talking about.21:14
ignaswell, now that i know that it's cache that is causing the failure i can fix it21:15
ignasthe test21:15
j1mso are you saying that after the registration, you are still getting a method-not-allowed error?21:15
ignasj1m: yes21:15
j1mI'm not at all convinced that this is a caching problem.21:16
ignasj1m: if i am removing all the queries before the one that is testing for method-not-allowed - the test passes21:16
ignaspublication.callObject(request, obj) gets the right view, and calls the POST method21:17
j1mIt would be very nice to get a test that demonstrates this.21:23
j1mObviously without the ST code.21:23
j1mI can't seem to reproduce this.21:24
ignasJ1m: i'll try21:25
*** zbir` is now known as zbir21:25
j1mwhat is the site structure?21:26
j1mIs this in the root site?21:26
j1mA subsite of the root site?21:26
mgedminin schooltool there's only one site -- the schooltool application object, which replaces the zope 3 root folder21:28
j1mIs it persistent?21:28
mgedminclass SchoolToolApplication(Persistent, sample.SampleContainer,21:29
mgedmin                            SiteManagerContainer):21:29
*** admp has quit IRC21:29
mgedminhm, I don't recall seeing this before:21:29
mgedmin    def __init__(self):21:29
mgedmin        super(SchoolToolApplication, self).__init__()21:29
mgedmin        notify(ApplicationInitializationEvent(self))21:29
j1mI assume that in your test, you create a new site.21:30
mgedminI suppose ApplicationInitializationEvent takes care to initialize the site21:30
mgedminor at least it used to21:30
j1mCould the test be loading the site from a database?21:30
j1mfor example, does the test run in a different transacton from the one that created the site?21:31
mgedminthat is very possible21:31
mgedminwait, no21:31
mgedminI'm suddenly confused21:31
mgedminit's a functional test21:31
*** Aiste has joined #zope3-dev21:31
mgedminIIRC ftests share the root folder and access it from many transactions21:31
mgedminand roll back any changes done between tests21:31
mgedminby manipulating ZODB storages21:32
mgedminso I guess the test does run in a different transaction from the one that created the site, yes21:32
j1mSo the root folder is created during functional test setup and then reread for each test.21:32
ignasi think so21:34
*** srichter has joined #zope3-dev21:34
j1mah, ok, I think I see the bug.21:35
j1mI'll look into it.21:35
j1mI'd rather you not "fix" your test.21:35
j1mbut wait for me to fix z3.21:35
mgedminwell, the test is not very nice either: it fiddles with the component architecture and doesn't clean up after itself21:36
mgedminit should instead register the extra view it needs from the appropriate foo-ftesting.zcml package include21:37
j1mYes. That's true.21:37
j1mI saw lots of tests like this.21:37
j1mIn the z3 sources.21:37
j1mThe new component-registry architecture should make it much easier to manage test-specific setup in the future.21:38
*** jukart has left #zope3-dev21:40
*** sashav has quit IRC21:43
*** sashav has joined #zope3-dev21:43
*** andres has quit IRC21:50
*** admp has joined #zope3-dev21:58
*** MJ has joined #zope3-dev22:00
*** ignas has quit IRC22:04
*** admp has quit IRC22:07
*** admp has joined #zope3-dev22:07
*** admp has quit IRC22:08
*** Egon has joined #zope3-dev22:10
*** ChanServ sets mode: +o srichter22:18
*** mkerrin has quit IRC22:19
*** dobee has quit IRC22:20
*** kamalgill has joined #zope3-dev22:25
*** alecm has joined #zope3-dev22:26
*** efrerich has quit IRC22:28
*** mgedmin has quit IRC22:36
*** alga has joined #zope3-dev22:38
*** alga has quit IRC22:38
*** roym has joined #zope3-dev22:41
*** rockyburt is now known as rockyburt|hockey22:52
*** Egon is now known as efrerich23:12
*** sashav has quit IRC23:28
*** sashav has joined #zope3-dev23:28
*** dobee has joined #zope3-dev23:30
*** MJ has quit IRC23:31
*** alecm has quit IRC23:33
*** regebro has quit IRC23:40
*** zbir has quit IRC23:48
*** benji has quit IRC23:52
*** sashav has quit IRC23:56
*** efrerich has quit IRC23:59

Generated by 2.15.1 by Marius Gedminas - find it at!