IRC log of #zope3-dev for Friday, 2007-11-16

*** rcrafton has joined #zope3-dev00:00
*** alga has joined #zope3-dev00:00
*** b52laptop has joined #zope3-dev00:00
*** salfield has joined #zope3-dev00:00
*** jsadjohnson has joined #zope3-dev00:00
*** niemeyer has joined #zope3-dev00:00
*** ccomb has joined #zope3-dev00:00
*** timte has joined #zope3-dev00:00
*** lmiller has joined #zope3-dev00:00
*** philiKON_ has joined #zope3-dev00:00
*** whit has joined #zope3-dev00:00
*** bigkevmcd has joined #zope3-dev00:00
*** instantf1o has joined #zope3-dev00:00
*** SteveA has joined #zope3-dev00:00
*** _Basti__ has joined #zope3-dev00:00
*** dokai has joined #zope3-dev00:00
*** nouri has joined #zope3-dev00:00
*** dobee has joined #zope3-dev00:00
*** RaFromBRC has joined #zope3-dev00:00
*** J1m has joined #zope3-dev00:00
*** run|home has joined #zope3-dev00:00
*** tomus has joined #zope3-dev00:00
*** Theuni has joined #zope3-dev00:00
*** WebMaven has joined #zope3-dev00:00
*** srichter has joined #zope3-dev00:00
*** Pan]_ has joined #zope3-dev00:00
*** romanofski has joined #zope3-dev00:00
*** dennis__ has joined #zope3-dev00:00
*** sets mode: +o srichter00:00
*** whit has quit IRC00:00
*** jhauser has quit IRC00:00
*** whit has joined #zope3-dev00:00
*** amaron has quit IRC00:06
*** norro has quit IRC00:09
*** rcrafton has quit IRC00:13
*** edgordon has joined #zope3-dev00:13
philiKON_srichter: why do we still need a package for the "zope3" kgs?00:14
philiKON_i don't understand the purpose of zope.kgs vs. zope.release00:14
philiKON_i thought zope.release was the software00:14
philiKON_and all you need to define a KGS is a controlled-packages.cfg00:15
philiKON_and a buildout.cfg00:15
srichterit also maintains the controlled-packages.cfg00:15
philiKON_that installs the zope.release eggs00:15
philiKON_why don't we split the controlled-packages.cfg away from zope.release?00:15
philiKON_zope.release should be the software00:15
philiKON_i don't think we need a *package* for controlled-packages.cfg00:15
srichterwell, that's what I am doing; just the other way around00:15
philiKON_we just need a location in svn00:16
philiKON_what you're doing sounds more complicated :)00:16
srichterI take pretty much all the non-Zope specific software away from zope.release00:16
philiKON_anything that starts with 'zope.' sounds like a package to me00:16
srichterzope.release will still contain the Zoe Tree fixing script though00:16
philiKON_what's zope specific in zope.release?00:16
philiKON_hmmm, i see00:16
srichtermaybe it won';t be a package; I dunno yet; I am just playing a little bit; I just had to check in so I won't loose the history00:17
* philiKON_ goes back to work :)00:17
*** Pan] has joined #zope3-dev00:34
mcdoncsrichter: fwiw, we're currently working on a project that can generate a kgs from a working set here:
philiKON_mcdonc: btw, when are you guys going to use reStructuredText, like the rest of the world? :)00:46
mcdoncsrichter: the intent is to turn the packages installed in the sandbox that a developer uses (and tests with) into a KGS00:46
mcdoncphiliKON_: when hell freezes over00:46
* philiKON_ feels like seeing zope 2.2 docs when he sees that indented text ;)00:46
mcdoncphiliKON_: i think in stx, unfortunately00:46
* philiKON_ should write an stx2rest converter00:47
mcdoncits ironic that python programmers are bothered by indentation00:47
philiKON_it's *really* hard to restructure stx00:47
mcdoncnot when you use a real editor ;-)00:47
philiKON_or you don't know how to use it ;)00:47
* philiKON_ has to go back to work00:47
mcdoncphiliKON_: yeh stop slacking off, ya slacker ;-)00:47
*** stub has quit IRC00:48
*** dobee has quit IRC00:50
*** Pan]_ has quit IRC00:51
*** norro_ has quit IRC00:52
*** lucielejard has quit IRC01:03
*** RaFromBRC has quit IRC01:07
*** RaFromBRC has joined #zope3-dev01:07
*** benji has quit IRC01:07
spythonhi, i have a custom widget with a template file which uses some 3rd party javascript magic to get the widget's input01:09
spythonthe problem is that the value of this field is put in the name "foo", but the widget expects something in the name ""01:09
spythoni don't have control over this javascript so i cannot rename it to use ""..01:10
philiKON_you  might be able to set the prefix of the widgets to ''01:11
spythoncan i force my widget not to use the "form" prefix in the name?01:11
spythonyeah, I'll try that but i wasn't sure that was a good way to do01:12
*** regebro has left #zope3-dev01:14
Pan]hehe. zope is funny in this way... i think that your idea is good, and few hours later, during solving some other problem and searching zope codes, you discover that there is much better way to do it ;]01:14
Pan]i mean "i think that my idea is good "01:15
Pan]ofcourse it is zope plus, that it has a lot of ready-to-use solutions. minus is that there is often no "designing practices"  documented.01:17
spythonhmm, form_fields['foo'].prefix = 'blah'   becomes '' :(01:21
spythoncannot touch the "form" prefix01:21
philiKON_no, i think you can set the prefix of the overall form01:22
philiKON_but isn't the javascript flexible enough so you can pass in the name of the form element?01:22
philiKON_sounds like some bad js...01:23
spythonhehe, it's the recaptcha javascript01:23
spythonyes i will try setting the overall form prefix..01:24
*** lmiller has quit IRC01:24
*** natea|boston has joined #zope3-dev01:26
*** rmarianski has quit IRC01:26
spythonyep it's working, cheers philip01:28
*** pcardune has joined #zope3-dev01:30
*** ccomb has left #zope3-dev01:30
*** dunny has joined #zope3-dev01:31
*** J1m has quit IRC01:32
*** RaFromBRC is now known as RaFromBRC|lunch01:35
*** greenman has joined #zope3-dev01:38
*** philiKON has joined #zope3-dev01:38
*** quodt has quit IRC01:42
*** b52laptop has quit IRC01:45
*** run|home has quit IRC01:48
*** philiKON_ has quit IRC01:54
*** pcardune has quit IRC01:57
*** TresEquis has quit IRC01:57
*** jsadjohnson has quit IRC02:01
*** BjornT has quit IRC02:02
*** pbugni has quit IRC02:09
*** spython has quit IRC02:10
*** RaFromBRC|lunch has quit IRC02:10
*** pcardune has joined #zope3-dev02:29
*** RaFromBRC has joined #zope3-dev02:32
*** mrevoir has joined #zope3-dev02:33
*** alecm_ has joined #zope3-dev02:39
*** hazmat has joined #zope3-dev03:05
*** salfield has quit IRC03:06
*** alecm has quit IRC03:09
*** mrevoir has quit IRC03:09
*** natea|boston has quit IRC03:09
*** natea|boston has joined #zope3-dev03:09
*** greenman has quit IRC03:11
*** huajie has joined #zope3-dev03:20
*** xenru has joined #zope3-dev03:25
*** tfws has joined #zope3-dev03:33
*** tfws is now known as darryl03:36
*** darryl is now known as darrylcousins03:36
*** Ariel_Calzada has joined #zope3-dev03:53
*** philiKON has quit IRC04:05
*** Ariel_Calzada has quit IRC04:05
*** whit has quit IRC04:11
*** projekt01 has joined #zope3-dev04:25
*** Pan]_ has joined #zope3-dev04:38
*** lucielejard has joined #zope3-dev04:38
*** pcardune has quit IRC04:42
*** alecm_ has quit IRC04:43
*** hazmat has quit IRC04:44
*** RaFromBRC has quit IRC04:46
*** dunny has quit IRC04:50
*** Pan] has quit IRC04:52
*** alga has quit IRC05:05
*** niemeyer has quit IRC05:09
*** dunny has joined #zope3-dev05:12
*** stub has joined #zope3-dev05:15
*** natea_ has joined #zope3-dev05:29
*** dunny has quit IRC05:35
*** natea|boston has quit IRC05:44
*** lucielejard has quit IRC05:47
*** dunny has joined #zope3-dev05:57
*** edgordon has quit IRC05:59
*** natea_ is now known as natea06:08
*** rcrafton has joined #zope3-dev06:11
*** pcardune has joined #zope3-dev06:21
*** cursor has joined #zope3-dev06:24
*** dunny has quit IRC06:25
*** lmiller has joined #zope3-dev06:49
lmillerI have a class that I have attached IAttributeAnnotatable to (via implements in the zcml), now I get a ForbiddenAttribute __annotations__ error. There's no adapter to set as trusted, as IAttributeAnnotable is added directly to the class. Any suggestions?06:50
pcardunelmiller: what is the particular line of code for which you get the error?  Something like IAnnotation(objectThatImplimentsAttributeAnnotatable)?06:57
*** baijum has joined #zope3-dev06:57
lmillerpcardune: I think the problem was I WAS using an adapter, I set it to trusted and it went on to my new error06:57
lmillerI find in zope that the errors can fly around a bit (all over the code)06:57
pcarduneyep.  security errors can be kind of tricky at times06:58
*** cursor has quit IRC07:04
*** natea has quit IRC07:19
*** natea|boston has joined #zope3-dev07:19
*** stub has quit IRC07:31
*** projekt01 has quit IRC08:13
lmillerthanks for your help08:23
*** lmiller has quit IRC08:23
*** afd_ has joined #zope3-dev08:25
*** jodok has quit IRC08:28
*** darrylcousins has quit IRC08:31
*** hdima has joined #zope3-dev08:33
*** dobee has joined #zope3-dev08:38
*** __mac__ has joined #zope3-dev08:57
*** afd_ has quit IRC09:07
*** sorin has joined #zope3-dev09:10
*** sorin is now known as sorindregan09:10
*** Jell-O-Fishi has quit IRC09:11
*** reco has joined #zope3-dev09:12
*** jodok has joined #zope3-dev09:21
*** markusleist has joined #zope3-dev09:22
*** philiKON has joined #zope3-dev09:31
*** schwendinger has joined #zope3-dev09:48
*** schwendinger has quit IRC09:49
*** jfroche has joined #zope3-dev09:50
*** malthe has joined #zope3-dev09:58
*** philiKON has quit IRC09:58
*** schwendinger has joined #zope3-dev10:06
*** gimni has joined #zope3-dev10:11
*** yvl has joined #zope3-dev10:11
*** greenman has joined #zope3-dev10:19
*** afd_ has joined #zope3-dev10:22
*** quodt has joined #zope3-dev10:30
* baijum just got his GSoC T-Shirt :)10:36
*** gstratton has joined #zope3-dev10:38
*** cursor has joined #zope3-dev10:44
*** agroszer has joined #zope3-dev10:52
*** harobed has joined #zope3-dev11:06
*** stub has joined #zope3-dev11:08
*** jodok has quit IRC11:28
*** norro has joined #zope3-dev11:34
*** b52laptop has joined #zope3-dev11:45
*** MJ has joined #zope3-dev11:47
*** pelle__ has joined #zope3-dev11:51
*** BjornT has joined #zope3-dev11:53
*** MJ is now known as MJ|out11:55
*** philiKON has joined #zope3-dev12:02
*** thruflo has joined #zope3-dev12:04
*** jodok has joined #zope3-dev12:09
*** greenman has quit IRC12:40
*** MJ|out has quit IRC12:43
*** MJ has joined #zope3-dev12:44
*** ktwilight has joined #zope3-dev12:54
*** ktwilight_ has quit IRC12:55
*** MJ is now known as MJ|lunch13:05
*** jsadjohnson has joined #zope3-dev13:09
*** toutpt has joined #zope3-dev13:13
*** pelle__ has quit IRC13:13
*** mkerrin has joined #zope3-dev13:16
*** ignas has joined #zope3-dev13:16
*** mgedmin has joined #zope3-dev13:24
*** ghendi has joined #zope3-dev13:26
*** thruflo has quit IRC13:28
*** thruflo has joined #zope3-dev13:29
*** cursor_ has joined #zope3-dev13:32
*** pelle__ has joined #zope3-dev13:33
*** cursor has quit IRC13:39
*** instantf1o is now known as instantfoo13:55
* mgedmin blinks14:00
*** MJ|lunch is now known as MJ14:00
*** jinty has joined #zope3-dev14:05
*** niemeyer has joined #zope3-dev14:11
*** afd_ has quit IRC14:30
*** pcardune has quit IRC14:30
*** markusleist has quit IRC14:31
*** duckinator has joined #zope3-dev14:33
*** baijum has quit IRC14:44
*** regebro has joined #zope3-dev15:04
*** __mac__ has quit IRC15:07
*** toutpt has quit IRC15:10
*** toutpt has joined #zope3-dev15:11
*** benji has joined #zope3-dev15:11
*** toutpt has quit IRC15:11
*** MJ has quit IRC15:23
*** projekt01 has joined #zope3-dev15:28
*** thruflo_ has joined #zope3-dev15:29
*** J1m has joined #zope3-dev15:30
*** thruflo has quit IRC15:44
*** Pan]_ has quit IRC15:47
*** pyqwer has joined #zope3-dev15:54
*** bigkevmcd has left #zope3-dev16:07
*** hexsprite has joined #zope3-dev16:12
*** jodok has quit IRC16:12
*** jodok has joined #zope3-dev16:13
*** jodok has joined #zope3-dev16:15
*** dobee has quit IRC16:15
*** jsadjohnson has quit IRC16:17
*** run|home has joined #zope3-dev16:26
*** lucielejard has joined #zope3-dev16:29
*** yvl has quit IRC16:39
*** jinty has quit IRC16:40
*** jinty has joined #zope3-dev16:40
*** afd_ has joined #zope3-dev16:41
*** toutpt has joined #zope3-dev16:53
*** whit has joined #zope3-dev17:09
*** whit has quit IRC17:10
*** dobee has joined #zope3-dev17:11
*** run|home has quit IRC17:13
*** whit has joined #zope3-dev17:14
*** rmarianski has joined #zope3-dev17:15
*** schwendinger has quit IRC17:16
*** dobee has quit IRC17:18
*** hdima has quit IRC17:20
*** ghendi has quit IRC17:20
*** sorindregan has quit IRC17:21
*** cursor has joined #zope3-dev17:25
*** baijum has joined #zope3-dev17:27
*** cursor_ has quit IRC17:30
*** thruflo_ has quit IRC17:37
*** thruflo has joined #zope3-dev17:37
*** thruflo has joined #zope3-dev17:37
*** natea_ has joined #zope3-dev17:39
*** huajie has quit IRC17:54
*** natea|boston has quit IRC17:55
*** jsadjohnson has joined #zope3-dev17:58
*** lucielejard_ has joined #zope3-dev18:00
*** quodt_ has joined #zope3-dev18:01
*** lucielejard has quit IRC18:07
*** jodok has quit IRC18:08
*** quodt has quit IRC18:10
*** baijum has quit IRC18:12
*** pcardune has joined #zope3-dev18:20
*** cursor has quit IRC18:20
*** natea_ is now known as natea18:21
*** lucielejard_ is now known as lucielejard18:21
srichterJ1m: any chance we will see a new ZODB release soon?18:34
srichterJ1m: well, good morning first of all :-)18:34
*** harobed has quit IRC18:35
J1mI am swamped, so I have almost no idea.18:35
J1mI know Theuni has been trying to resolve critical bugs.18:35
srichteryeah, he fixed the finel new critical bug18:36
srichterplus some more18:36
J1mI haven't had time to review them, so I hope they're good. :)18:36
srichterhe had a question about the solution on one of them18:36
J1mHm, I guess I missed that.18:37
srichterno, you responded I just saw18:38
mcdoncbefore there is a new ZODB release, we should fix this too..
J1mmcdonc, we're talking about 3.8, which shouldn't be affected by that.18:39
mcdoncJ1m: ok18:39
srichtermcdonc: but I thought that was on the trunk18:39
srichtermcdonc: ZODB 3.8 has alreadya tag as far as I know18:39
mcdoncsorry, i didnt know you were talking about 3.8 (you have esp with each other ;-) )18:40
srichtermcdonc: I just had annoyed jim enough about it before18:40
*** pcardune has quit IRC18:40
srichterJ1m: so 98275 is the last remaining critical bug altogether18:41
srichtermcdonc: do you want power? Do you want access to all Zope packages on PyPI that I have access to? :-)18:44
mcdoncsrichter: no, but i'll take it ;-)18:44
* J1m resists pointing out on distutils-sig how absolutely self destructive it is to use system Pythons.18:47
* J1m feels better for venting somewhere. :)18:47
ignasJ1m: please link18:47
ignasoh, pointing out18:47
ignasi thought there was a post in which someone was explaining why it can be disastrous and how18:48
mgedminthere were multiple mailing list threads, well, at least one18:48
srichterJ1m: poor benji, got all the heat18:48
J1mbenji likes heat. Otherwise, he gets chilly.18:49
*** rmarianski has quit IRC18:49
* benji shivers, due to lack of heat.18:49
* mcdonc shakes his head at people who are amazed by the default python leopard install ("it has twisted in it already!")18:50
*** gstratton_ has joined #zope3-dev18:50
*** pelle__ has quit IRC18:51
*** TresEquis has joined #zope3-dev18:52
ignasJ1m: hmm, not a very good example, I had the same problem when *compiling* my own python, because i didn't install libzlib-dev first18:54
*** gstratton__ has joined #zope3-dev18:59
srichtermcdonc: all done; do you feel th force? :-018:59
mcdoncsrichter: if by "do you feel the force" you mean "am i crying", then yes ;-)18:59
ignassrichter: can i have some too, so i could release "bugfixes" that i am too lazy to write an email about to zope-dev ;)19:00
* srichter wants PyPI groups really bad19:00
*** projekt01 has quit IRC19:00
*** nathany has joined #zope3-dev19:01
srichterignas: I guess you know the drill? update CHANGES.txt, create release tag, bump version, etc?19:01
ignasnot really, though i had some practice with some schooltool eggs following steps described by philiKON in some text file about zope release process19:02
ignasmaybe without the changelog part19:03
srichterusername on PyPI?19:03
ignaslet me check19:03
srichter(I think it is good if at least one person of each major project or company has access to those packages19:03
*** alecm has joined #zope3-dev19:03
ignas"ignas" seems to be the username19:04
srichterit's rolling now19:04
*** gstratton has quit IRC19:05
*** mcdonc has quit IRC19:10
srichterignas: you also got the force, young ignas19:12
ignassrichter: thanks19:12
*** philiKON has quit IRC19:12
ignasoooh, my list of packages is LONG19:12
J1mignas, I in this example, the user had zlib installed.  I agree though that, in this case, the underlying issue is that the zlib module should no-longer be considered optional in Python.19:13
*** natea has quit IRC19:13
srichterJ1m: I think this will cause a lot of resistence, because setuptools in itself is not a Python pacakge19:14
J1mI don't think so. Guido has stated that he wants setuptools in the standard lib.  And zip import is a fairly core feature of Python.19:15
* mgedmin hopes setuptools will eventually become sane, stable and part of stdlib19:15
*** gstratton_ has quit IRC19:16
TheuniJ1m: any change to look at the zodb bug? I'll be gone for two weeks starting tomorrow, just to let you know ...19:16
Theunion another topic: does anybody know why the publisher strips away trailing whitespace from the PATH_INFO?19:16
Theunithere is no test case that verifies that it does this and IMHO this is wrong.19:17
srichterTheuni: I would try to see what happens if it does not do this by running all the tests19:18
TheuniI haven't played with your tools yet, I guess running the tests from the old tree would be sufficient.19:19
Theunis/any change/any chance/19:19
srichteryeah, that should be sufficient19:20
Theunii mean. even if it breaks something, that other thing would be wrong as well ;)19:20
srichtertesting against the KGS is trivial though; just downlaod
Theuniwhat currently happens is that urls like /foo%20 are turned into /foo19:21
srichterahh, that does sounds like a mistake19:21
Theuniand it's been there for a *looooong* time19:21
srichterI guess people just do not have this use case19:22
Theunii just had19:22
Theunibecause a customer entered something where he accidentally a space at the end of something that was being used as a name19:22
Theuni(insert your favorite missing verb)19:23
Theuniit might be a bad idea to allow him doing that anyway, OTOH the publisher is broken in that regard19:24
*** philiKON has joined #zope3-dev19:29
*** mcdonc has joined #zope3-dev19:31
ignassrichter: if i want to make a backwards incompatible change - should i just bump the version up like 3.4->3.5 , make a branch for 3.5 that only has dev releases or something else?19:37
ignasi made a mistake when doing the signature of "startServer" function in zope.testing.server19:37
ignasand would like to change it19:37
philiKONcan't you just create a new function with the improved signature?19:38
ignasstartImprovedServer ;)19:38
ignasthough i guess startHttpServer will be good enough19:38
ignaswith a deprecation waring in the old function19:39
*** quodt_ has quit IRC19:41
TresEquisTheuni: can we nag you as a zagy proxy?19:42
TresEquisOh, I see that zagy is here too19:42
Theuniyou can nag zagy as a proxy for me =)19:42
Theunii'm almost gone to my vavation :)19:42
Theunivacation indeed19:42
Theunizagy just left heading home, i'm not sure he'll be online in the next hours19:42
TresEquiswe need to get non-borked eggs released for the .dev-r### eggs currently on the cheeseshop19:42
Theuniwhat's up?19:42
TresEquisI figure he knows what was the motivation for releasing them19:43
Theuniwanna do it yourself?19:43
Theuniah ok19:43
TresEquisbecause he released most of them ;)19:43
Theunii'd write him a mail19:44
Theunii'm currently wrapping up my stuff and really don't wanna open up anything new :)19:44
timteI have problems using zope3 widgets in zope2, looks like it's because zope2 uses content-type and zope3 uses Content-Type19:45
TresEquisis it
*** zagy has quit IRC19:49
TresEquis"cubic zirconium" -)19:49
mcdoncTheuni: alternately we could delete the 3.5.0a2 release, if the g'damn cheeseshop let us.19:49
TresEquissomebody would whine, but maybe it would be the "right" people19:50
Theuniwell, the cheeseshop does let you deleete19:50
TresEquishmm, we can't even see that release on the non-simple page19:50
Theunithat's because it's hidden19:50
Theuniare you logged in?19:51
srichterTresEquis: Theuni: well, he added layer support for XMLRPC19:52
Theunithat page, at the end lists all versions for me19:52
Theuniincluding a 'remove' button19:52
TresEquissrichter: OK, but hee needs to finish the release19:52
Theuni(note that i'm not advocating anything. i'm just pointing out the mechanics *g)19:52
TresEquisor somebody does19:52
*** gimni has quit IRC19:53
* mcdonc stashes that url away ;-)19:53
*** pigeonflight has joined #zope3-dev19:53
mcdoncThthank you19:53
mcdoncTheuni: thank you19:53
srichterTresEquis: Theuni: that feature was hotly contested and should be removed19:53
TresEquisthe presence of that egg on the simple page is a thorn19:53
srichterI would prefer removing the release and revert the svn19:56
srichterwhich would be probably as easy as reverting back to the checkin before the changes, because not much has happened with those packages.19:56
J1mTheuni, I'm unsure what the status of is.19:56
J1mOr even of the details of the problem.19:57
Theunithe status is that i have two proposed solutions available19:57
mcdoncsrichter: i just blew 3.5.0a1 and 3.5.0a2 away and i'm testing installing zope.tal now19:57
Theunithe details are that if an object is modified during __getstate__ in a ZEO context that the cache is tried to be invalidated for a _current_ version of an object which is guarded by an assertion right now19:58
*** pigeonflight is now known as pigeonout19:58
J1mTheuni, look at the dialog I tried to open to you.19:59
Theuniah darn. i'm not logged in19:59
*** schwendinger has joined #zope3-dev20:00
*** jfroche has quit IRC20:03
mcdoncsrichter: this fixed the dependency chain for at least zope.tal..we'll see if if fixes all the others when the bot runs tonight20:03
mcdonc(at least for some definition of "fixed".. people who install from cheeseshop will likely have a wildly different "zope" than they would if they installed from the kgs)20:03
srichtermcdonc: at this point the zope.* and* releases are not very diverged, so it should be pretty okay20:04
*** afd_ has quit IRC20:04
srichtermcdonc: I hope to start a dev-kgs soon that will hopfully be very close to what the latest version in cheeseshop is20:04
*** romanofski has quit IRC20:06
mcdoncsrichter: cool.  have you had a chance to look at my changes to zope.tal in order to approve a new minor release?20:06
*** reco has quit IRC20:07
*** quodt has joined #zope3-dev20:07
*** pcardune has joined #zope3-dev20:07
srichtermcdonc: I did not get your checkin message20:07
srichtermcdonc: are you subscribed to
mcdoncwhich list would it have went to?20:10
srichterwhich includes the entire SVN20:10
srichterI think you change is fine, because the dummy engine is never looked up in the CA20:10
mcdoncthats what i thought too, very good20:11
srichterI wonder whether TAL should carry its own ITranslationDomain interface...20:11
ignassrichter: what do you mean?20:12
mcdonci didnt get the checkin message either, fwiw20:12
srichterwell, TAL has support for translating20:12
mcdoncare there archives for that list?20:12
srichterand in order to hook a translation "thing" into it it must fulfill some API20:12
srichtermcdonc: I dunno; I am using the web page now to review the diffs20:12
mcdoncphiliKON: that was zodb... i made a separate checkin to zope.tal the other day20:14
srichtermcdonc: I think you are good; it would be great to see a 3.4.1 release of zope.tal in this case.20:14
*** schwendinger has quit IRC20:14
mcdoncsrichter: is it just a matter of me doing a release to the cheeseshop?  do we put it in the KGS?  do we make a new KGS?20:15
*** pigeonout has quit IRC20:15
srichtermcdonc: yeah, I can do the KGS20:16
srichtermcdonc: releasing on cheeseshop and updating the KGS are not coupled yet20:16
srichterwhich I think is the point of the KGS in the first place ;-)20:16
philiKONmcdonc: i wonder if zope.testing can grok a utility function for the 'alltests()' business you're doing in
mcdoncphiliKON: +120:17
J1mI keep threatening to implement a meta-index that routs uploads to multiple indexes.20:17
srichterI think even more important would be to switch to using the test_require option in buildout20:17
srichterJ1m: that is what repose does in a way20:18
srichterJ1m: that is what repoze does in a way20:18
srichterJ1m: I take that back; I misread20:18
J1mok, then I didn't. :)20:18
ignashmm, i'd think uploading to KGS needs different testing than a release to PyPI20:18
srichterit reroutes downloads20:18
srichterignas: it does; you have to test in the set of the KGS, which is easy to do20:19
srichterthe scripts are all setup in zope.release20:19
srichterJ1m: but the idea is good :-) it would allow us simulating a zope group for PyPI20:20
srichterJ1m: I wanted to do that too, but I dismissed it as too much work with too little benefit20:21
mcdoncsrichter: not repoze, but compoze.  it allows downloads from multiple indexes but only allows creation of a single index, though.20:22
*** jodok has joined #zope3-dev20:28
mcdonci think we're also going to try to make a catalog client that is willing to consult multiple indexes to satisfy dependencies.. i think phillip eby might have already started something like this though (at least i heard a rumor of something called 'neat' or.. something)20:29
philiKON nest20:30
mcdoncnest.  right.  that's what i said. ;-)20:31
TresEquisphiliKON: can take a 'test_loader' argument,  in addition to 'test_suite' or 'test_module'20:32
TresEquisI'm going to need to release a new 'eggtestinfo' now20:32
philiKONso the new version of eggtestinfo can plug zope.testing's test runner into test?20:33
J1mNo, it will leave turds behind that the testrunner recipe can be trained to pick up. :)20:33
philiKONwell, that's what it currently does, right?20:34
J1mwhat "it" are you refering to?20:34
J1mTresEquis, we'll have to tell people how to use this plugin.20:34
J1mphiliKON, yes20:34
J1mbut the testrunner recipe will have to be updated to look for this meta data.20:35
J1mas it is new meta data20:35
srichterthat would be really great to have, indeed20:36
*** mgedmin has quit IRC20:36
J1mabout as great as text extras. :)20:36
srichterJ1m: yes, except that it plays nicer with the setuptools-only world20:36
srichterJ1m: I would hate to need to specify both, test_require and extra_requires->test20:37
J1mso don't specify either.20:37
J1mBut I support this effort -- don't get me wrong.20:38
*** mcdonc_ has joined #zope3-dev20:39
srichterJ1m: I know you want it all in install_reqires, though I think that people have come up with very good examples why you would not want to do this.20:40
J1mI think Deiter had a reasonable example.20:40
J1mI'm tired of arguing over and over about these so I didn't bother arguing with the others.20:40
philiKONi do support benji's law, but if this is a way to practically reduce our ridiculous system of interpdenencies, i'm all for practiciality20:41
J1mI don't think it is.20:41
TresEquiseggtestinfo just modifies the 'egg_info' comman20:41
srichterwhat is Benji's Law? Nailing all versions?20:42
philiKONtest what you fly and fly what you test20:42
TresEquisit adds data about the testing stuff declared in to a .egg-info/test_info.txt file20:42
srichterTresEquis: so this has to be installed at what point?20:42
srichterI guess we would have to update
J1mTresEquis, I assume someone has to add something to their to get this extension.20:42
TresEquiswhen you run ' sdist'20:42
philiKONTresEquis: uh, when you run egg_info, right?20:42
TresEquisor ' egg_info' or ' bdist*'20:42
philiKONwhich buildout does20:42
philiKONsrichter: i think there's setup_requires?20:43
TresEquissdist and bdist* all run egg_info behind the scenes20:43
J1msrichter, why would we have to update
philiKONwe don't20:43
philiKONthere's setup_requires20:43
TresEquis' test' runs it too20:43
srichterJ1m: I did not know about setup_requires, *sigh*20:43
*** regebro is now known as regebro|food20:44
J1msrichter, think of setuptools as a long journy of discovery -- like zope. :)20:44
*** RaFromBRC has joined #zope3-dev20:44
philiKONpje *did* write that zope was an inspiration to his work on wsgi and setuptools :)20:45
srichterJ1m: he he. I can tell. Geez. It's just a packaging tool20:45
* J1m worries that pje tool wwzd too far20:45
mcdonc_if we always did test what you fly, fly what you test, we'd be need to be flying around in this thing.. ;-)20:45
philiKONwell, that's where the moon landing really took place, right? :)20:46
*** mkerrin has quit IRC20:46
*** toutpt has quit IRC20:46
mcdonc_i think we might be able to streamline that building a little.. take some edges off the corners... then ship it20:47
*** mcdonc has quit IRC20:49
*** afd_ has joined #zope3-dev20:49
mcdonc_if we can manage some tests_require analogue in buildout, we can unhose the current (very conservative) zope.i18n requirements.  alternately, we can just change the software.20:53
*** natea|boston has joined #zope3-dev20:54
mcdonc_i mean, right now i can "fix" zope.i18n's requirements in but buildout will choke20:54
mcdonc_(as it depends on install_requires)20:54
philiKONwhat exaclty did you want to remove from zope.i18n's install_requires?20:54
philiKONit's tiny compared to others :)20:54
philiKONwhat we could do is move out the zcml handling stuff20:55
*** cursor has joined #zope3-dev20:55
philiKONthat would dump the requirement on zope.configuration, zope.component20:55
mcdonc_philiKON: see
TresEquisphiliKON: exactly20:55
TresEquisZCML processing is the elephant in the living room here20:56
philiKONmcdonc_: but the changes you're proposing are also lies :)20:56
mcdonc_philiKON: they're more truthful i think20:56
philiKONTresEquis: agreed. i'm mostly to blame for the zope.* packages having zcml processing stuff :(20:56
TresEquisthe package itslelf doesn't need to *process* ZCML20:56
philiKONbut if somebody depends on zope.i18n for the zcml stuff ...20:56
TresEquisit just exports some for consumption by the framework20:57
philiKONright. how's that differnet than exporting an import?20:57
TresEquisthey should have the ZCML-processing eggs themselves as dependencies20:57
philiKONyes, in an ideal world20:57
TresEquisso they shouldn't be getting them as transitive depencences of an egg which doesn't itself actually use ZCML20:57
TresEquisIf somebody breaks because an egg which never imports zope.configuration quits depending on it,20:58
TresEquisthen we just fix them20:58
TresEquiswhich is what grok just did20:58
TresEquisit made its "transitive closure" explicit20:58
philiKONzope.i18n should depend on zope.zcmlfori18n?20:58
philiKONok, good :)20:59
TresEquiszope.i18n shouldn't depend on ZCML at all20:59
TresEquissomebody who wants to use ZCML from zope.i18n would need to pull in the zope.configuration[zcml] requirement20:59
srichterwell, that's not necessarily true; zope.i18n was originally written for Zope 3, which means support for ZCML is intended20:59
srichterthis cannot be changed that easily21:00
TresEquisbut it doesn't *need* ZCML21:00
srichterbecause it would effectively break all packages21:00
TresEquisZope3 already *has* ZCML processing eggs21:00
srichteroh, but who uses that?21:00
srichtercome on21:00
TresEquisyou don't need to do ZCML to use the zope.i18n egg21:00
philiKONsrichter: um... we?21:01
srichterI don't21:01
TresEquisyou only need zope.configuration[zcml] if you want to load the ZCML which is inside the zope.i18n egg21:01
*** gimni has joined #zope3-dev21:01
srichterbecause pulls in too much stuff21:01
philiKONsrichter: that doesn't mean nobody uses it :)21:01
TresEquisbut if you want to load its ZCML, you are going to have the eggs already21:01
srichterI use it for the Zope 3 source tree, but that's it21:01
mcdonc_the only thing that depends on the zope.i18n.zcml module is the zcml itself and the tests21:01
philiKONit's functionality that it exposes21:01
philiKONi don't think that's wrong per se21:01
philiKONand that functionality is tested21:02
philiKONi don't see a logical problem here21:02
srichterphiliKON: I was irnoic about that trying to point out that if you use the eggs approach fully, you would not use
philiKONit's just that zcml support would be better off outside those eggs21:02
srichterZCML is a feature of zope.i18n21:02
srichterwith any feature in a package that requires a few more deps you can make the same argument21:03
srichterand I agree with philiKON, whether ZCML should be a feature of zope.i18n or not is a different question21:03
philiKONwasn't that why we split up zope in a gazillion eggs? :)21:03
srichteryes, which is the reason I was not happy about it21:04
philiKONi wonder if it had been such a problem if we had split up Zope only to about 5-10 eggs21:04
srichterthere is a real cost associated with maintaining a separate package. I think that cost has been evalauted too low by most21:04
srichterphiliKON: no, it would have not, because the dependency tree would have been much easier21:05
philiKONi mean, zope is still flexible enough that it doesn't hurt if you don't use it all21:05
srichteralso, it would then be no problem to sync the releases a little bit more, creating new packages for all 5-10 eggs whenever versions change21:05
TresEquistesting the loadability of the ZCML within the zope.i18n package tests imposes unreasonable costs on people who might use the egg, without wanting to do any ZCML at all.21:06
TresEquisat least if we move the requirement from 'install_requires' to 'tests_require'21:06
srichterTresEquis: sure, but that's a historical fact21:06
TresEquisthose people will be helped21:06
philiKONit's still a lie IMO :)21:06
TresEquisbecause they mostly won't run the tests21:06
*** romanofski has joined #zope3-dev21:07
TresEquisYou could make a case for a zope.i18n[zcml] extra21:07
srichterI tend to agree with philiKON, but I have to think about this much harder21:07
TresEquisbut Il Papa doesn't like extras21:07
srichterextras should not be used21:07
srichterread the archives21:07
TresEquisI never bought those arguments21:07
srichterJim has made that argument on the mailing list before21:07
TresEquisthe other logical choice is to split the egg21:07
TresEquiswhich is *more* expensive and painful21:08
srichterand I do agree with him; using the test extra was a big exception21:08
TresEquisagain, if I want to use the egg as a library, and never intend to load / use its ZCML21:08
*** timte has quit IRC21:08
TresEquisI shouldn't need the ZCML requirements21:08
*** afd_ has quit IRC21:08
TresEquisthat is the use case for which the extras feature exists21:08
*** rmarianski_ has joined #zope3-dev21:09
TresEquisI know J1m doesn't like them21:09
*** ignas has quit IRC21:09
TresEquisbut given packages which satisfy disjoint use cases, I don't see an alternative21:09
*** mcdonc has joined #zope3-dev21:09
srichterTresEquis: I understand thw use case you have perfectly; I am not sure about a solution yet21:09
*** afd_ has joined #zope3-dev21:09
TresEquis1) split the eggs21:10
srichterTresEquis: I also value testability above everything, so I agree with Jim that extras are harmful21:10
TresEquis2) use extras (e.g., a 'zcml' or 'tests' extra)21:10
TresEquis3) use 'tests_require'21:10
* J1m prefers to split the egg21:10
TresEquis2) is most precise in intent21:10
* J1m doesn't agree21:10
TresEquisok, 2 is more precise than 321:10
TresEquisbecause there *is* code in eggs which expose 'meta' ZCML which imports zope.configuration stuff21:11
srichterthe question is: for how many packages is such a split desirtable21:11
*** rmarianski_ is now known as rmarianski21:11
srichterif it is just for 10-20, I would prefer the overhead of extra packages over the testing problem21:11
J1mI Martins confusion is an excellent example of extras being missleading.21:11
TresEquis"I Martins"?21:12
J1mOr, more precisly, having code in the package that diesn't work unless some special chant is given at install time is confusing.21:12
philiKONTresEquis: insert "think" where appropriate :)21:12
J1mI think Martins confusion is an excellent example of extras being missleading.21:12
* srichter translates Jimisms: I Martins -> I think Martin's21:13
J1mThe "Zope without Zope" thread.21:13
* srichter should write a Jim translator bot21:13
J1mHe easy_installed zope.component and expected to use the zcml support that was included. Silly him.21:13
srichterhe he21:14
*** greenman has joined #zope3-dev21:14
J1m(well, he used buildout, which is beside the point)21:14
srichterTresEquis: do you know for how many packages you would want to separate the ZCML (in whatever way)?21:14
TresEquisI would prefer that all configure.zcml be moved into zope.app21:15
TresEquismeta.zcml could stay in the library packages21:15
philiKONundoing what we did in 3.2->3.3 :)21:15
mcdoncthere are currently ~ 200 eggs.  i think it'd be reasonable to use the "split the egg" pattern for things in zope.* (putting zcml support maybe in some package), but i think that at that point needs to be fewer eggs21:15
TresEquisbut I would prefer to have it *not* force people who don't want to use ZCML to install those dependencies21:15
TresEquisso, by J1m's logic, that would be "split eggs with meta.zcml"21:16
srichtermcdonc: I know the overall numbers, but specifcs of the eggs that want to be used as libs would be nice21:16
J1mI don't follow you TresEquis.21:16
TresEquissrichter: any egg outside should aspire to be libraryish21:17
J1mMy logic is split eggs so install-requires is reasonable.21:17
TresEquisJ1m: "libraryish" eggs shouldn't export configure.zcml21:17
TresEquisbut some export "useful" meta.zcml21:17
TresEquisIn that case, the package should either split (to keep depenencies minimal for the libraryish part)21:17
J1mzope.component doesn't export either.21:17
TresEquisor move the ZCML dependencies into an extra21:17
mcdoncsrichter: one strategy would be to strive for anything *not* in to "be straight libraries" (export no zcml), but thats probably impossible right now for b/c reasons21:17
J1mIt gets into trouble with meta.zcml.21:18
srichterTresEquis: can we stop talking about theoretical things? I know what would be ideal, but this is not an ideal world21:18
J1mI think it depends a lot on the intended use of the package.21:18
TresEquisI'm proposing real world changes21:18
*** mcdonc_ has quit IRC21:18
TresEquisany pacakge in zope.[!app] which currently depends on zope.configuration should be fixed21:18
mcdoncthis problem is completely solveable in packaging21:18
TresEquisby splitting or by adding extras21:18
mcdoncwithout changing any code21:19
TresEquisexcept BBB imports,21:19
TresEquisin the case of split packages21:19
*** J1m has left #zope3-dev21:19
srichterBBB will be tough with whatever solution we come up with, mmh21:21
TresEquismostly, we'll need to fix or supply bbb for the <include package ="" filename="meta.zcml" /> folks21:22
srichterthat's not all21:22
srichterpeople expect to say: require zope.i18n and get all the meta and real ZCML configuration with it21:23
srichterwe would definitely see no benefit of the work, till the BBB expires21:24
mcdoncwhat if we didnt split packages, and just changed install_requires to not lie, and made people that wanted zcml install zope.configuration [zcml]21:24
srichterthat's not enough21:24
srichterbecause people already use a simple zope.i18n specification in their projects21:25
TresEquisHow can people expect 'zope.i18n' to get the whole ZCML-processing engine ?21:25
srichterrequiring zope.i18n[zcml] is not BBB21:25
*** agroszer_ has joined #zope3-dev21:25
TresEquisif they want to process ZCML, their application has to kick it off, no?21:25
TresEquiswhich means it needs zope.configuration[zcml] already21:25
mcdoncsrichter: but it's a way, way, way simpler fix to tell them to install another egg too than touching *every* package21:25
TresEquisZCML isn't (yet :) majiykally processed21:25
TresEquisthank the gods21:26
mcdoncsorry s/touching every package/splitting every package21:26
TresEquiswhat appliacation would break if zope.i18n didn't say it depended on zope.configuration[zcml]?21:26
srichterwell, we first have to determine whether this is desirable for all packages21:26
TresEquisNot grok, not the Z3 meta-egg21:26
TresEquis(the latter may be only notional at the moment:  think of it has having the same dependencies as the latest versions in your controlled-packages.cfg)21:27
srichterokay, so install_requires now lists all packages needed in the ZCML21:27
mcdonci dont think we can determine that until we concretely figure out whether it works for every package before figuring out whether it will work for e.g. zope.i18n only21:27
srichterthis is often more than just zope.configuration21:27
TresEquisany "library-ish" package which contains non-meta ZCML21:28
srichterthe more I think about it, the worse it gets21:28
TresEquisand where that ZCML configures using packages which would otherwise not be dependencies of the package21:29
TresEquisis broken ;)21:29
srichterwell, that's life21:29
TresEquisit is really an application, and not a library21:29
srichterwe need to deal with this21:29
srichterZope 3 was not written with eggs in mind21:29
srichterI disagree21:29
TresEquisso we should split the "application" bits out from the "library" bits21:29
mcdonc"fixing" it and testing in in the context of the "major" top level apps (zope3, grok) is one day's work... other folks who have a  top-level dependency on 'zope.i18n' (of probably whom there are exactly zero) will need to add another dependency extra to their requirement specifications21:29
TresEquislibraries dont impose "policy-laden" dependencies on their clients21:30
TresEquisapplications *are* the clients, and should feel free to impose policy21:30
srichterit's not that I disagree with your definitions, but it deos nto reflect our situation21:31
*** schwendinger has joined #zope3-dev21:31
srichteranyways, at this point a mail to the zope-dev mailing list is in order21:32
TresEquissrichter: I'm arguing that eggs which are outside the namespace should be mutated to conform with this definition21:32
srichterI think we have determined all the options, none of which is perfect, so we should get feedback from others21:33
TresEquissplitting the library bits from the app bits will improve their usability21:33
*** alga has joined #zope3-dev21:33
TresEquisespecially for people who *aren't* already driinking our koolaid21:33
srichterI agree, but a lot of the Zope 3 developers have been burned by over-zealous decisions regarding eggs in the past year and became much more cautious21:34
TresEquisa lot of the burn comes from older decisions21:34
srichterwe barely are getting to point were things seem to work; most of us are very tired of interruptions21:34
TresEquise.g. the choice not to manage dependencies before we even thought of eggs21:35
TresEquis"anything in can depend on anything else in" was expedient at the time21:35
TresEquisbut *really* painful now21:35
srichterwe all know that21:36
TresEquisso I'm arguing for non-expedient investment *now*21:36
TresEquisto avoid compound interest later21:36
*** thruflo has left #zope3-dev21:37
srichterwe all know that too; the fact that you got input on IRC from so many people is a recognition that this is an issue we all see21:37
TresEquisyou've been doing some of that investment effort already21:38
srichterbut several investments in the last 6 months promised improvements that did not happen21:38
*** agroszer has quit IRC21:38
*** pelle__ has joined #zope3-dev21:38
TresEquiswhich I think we all appreciate21:38
TresEquisInvestment does imply risk21:38
TresEquissometimes you take a bath21:38
TresEquisbut we still learned from the failures21:39
TresEquis(at least, we hope so)21:39
srichterright, and if things go wrong too mcuh, you get burned, which we are now; so new investment oppurtuinities will undergo much more scrutiny21:39
*** mcdonc_ has joined #zope3-dev21:39
srichterJim has stated multiple times that we should move much slower now and I think there was silent consent21:40
TresEquisfixing our dependency story will help on multiple fronts21:40
mcdonc_why dont we just remove the eggs from the cheeseshop21:40
TresEquisit will make the eggs themselves smaller, cleaner, easier to maintain21:41
mcdonc_then it wont be an "emergency"21:41
TresEquisit will keep related policy together21:41
srichterbut people depend on cheeseshop already21:41
mcdonc_right now, people are tempted to actually use them ;-)21:41
TresEquisand away from reusable mechanism21:41
srichterpeople have deployed applications21:41
mcdonc_they can't use the kgs?21:41
TresEquispeople can change their configurations to point at your KGS+cheeseshop index21:41
srichtermcdonc_: the kgs does only provide links21:42
TresEquisLet's put the eggs somewhere we control21:42
srichtermcdonc_: it also is not a fully official story yet21:42
mcdonc_srichter: both solveable without making "hard" decisions21:42
srichteranyways, you need to bring up your ideas on the zope-dev mailing list before moving21:42
*** greenman has quit IRC21:43
* TresEquis needs to run21:43
*** TresEquis has left #zope3-dev21:43
srichterfor me right now, the status quo is better than anything else; I will consider proposed changes, but think hard about them; but then I am only one developer in the entire community21:43
mcdonc_me too...  sorry... i'll be back21:43
*** mcdonc_ has quit IRC21:43
*** mcdonc has quit IRC21:49
*** mcdonc has joined #zope3-dev21:56
*** afd_ has quit IRC22:03
mcdoncsrichter: sorry bout that... we've had a lot of conversation that piled on a whole lot of context that would be very difficult to give to impart to everybody..  if we did ask a question, it would probably need to be on the normal zope list, and i believe it would be need to be framed in the context of "we want to remove all zope-related distributions from the cheeseshop except for zope.interface, ZODB, and zope.component so we can work ou22:04
srichteryou broke up ;-)22:05
*** jodok has quit IRC22:05
srichterI think removal is not an option at this point22:05
mcdoncsrichter: well, then, we're gonna need to stay in "high pressure" mode to fix this stuff22:06
srichterI think you should make more feasible offers, such as sorting out the ZCML dependencies22:06
srichteranything very interuptive right now will cause a lot of resistence, I think22:06
mcdoncI am talking about sorting out the ZCML dependencies actually as part of "working out some packaging issues"22:07
mcdoncsrichter: given that the KGS works, and that no one has been able to actually install anything except the distros i excluded from the cheeseshop, i suspect it's a perfect time to do it22:07
*** rcrafton has quit IRC22:08
*** jodok has joined #zope3-dev22:08
mcdoncs/from the cheeseshop/using the cheeseshop as an index with non-buildout clients22:08
*** rcrafton has joined #zope3-dev22:08
srichtermcdonc: remove the packages you mean?22:09
srichtergood luck22:09
mcdoncok, not a good start then ;-)22:09
srichterI do not represent the entire community, but -1 from me ;-)22:10
mcdoncso... given that they stay up there... what do you propose to do22:10
srichterI think resolving the dependencies is a high priority22:10
srichterand people agree with you and Tres on this22:10
srichterwe have to think hard on how to do it22:10
*** rcrafton has quit IRC22:10
srichterTres summarized the three options already22:10
mcdoncok, well.. fixing the dependencies means releasing new eggs22:11
srichterI think they are all valid; we just have to decide the pros and cons of each22:11
srichtereach has they advantages and disadvantages22:11
mcdoncill note that while we're deciding, we look awful bad22:11
srichterI think that's okay though22:12
mcdoncbut maybe i just need to get over that ;-)22:12
srichterright now we are stabalizing the story for the people already using the Zope 3 stack22:12
mcdoncwhatever you call it, we're lookin bad ;-)22:12
srichterthat is more important to me right now than trying to attract new people22:12
srichterbut we are also early adopters22:13
mcdoncwhich is meaningless to the people who try to use it22:13
mcdoncbut anyway22:13
mcdoncok, given that the cheeseshop releases stay22:13
mcdoncwe're going to need some way to test new eggs... i think a private index is the best thing22:14
srichterI think that we should concentrate on the packages that are most likely to be used outside the Zope world, such as z.interface, z.component, z.tal, z.tales, z.pagetemplate22:14
mcdoncmaybe even local to our own machines22:14
mcdoncthat's fine, all those distributions will work once I release a new zope.tal package22:15
mcdoncbut i'll note that doing that is something that you've argued against today...22:15
benjiI've not really been following the conversation, but this idea just struck me: perhaps we should treat the cheeseshop as a KGS as far as the packages we manage22:15
mcdonc"someone might be depending on zope.tal to suck down"22:16
srichterI think it is impossible to treat the cheeseshop as KGS22:16
srichterthat would be the same as saying: the entire Zope trunk works in itself22:17
srichtersomething that we never achieved22:17
srichterone package was always outdated22:17
mcdoncbenji: we'd be able to do that if we only made the promise that "installing one meta egg from the cheeseshop will get you all these version-pinned things" or (the current story) "you need to use buildout"22:17
mcdoncthe individual subparts will have dependencies that cant be satisfied by the cheeseshop as a KGS22:17
benjiok; ignore me then ;)22:18
mcdoncbenji: i for one refuse to ignore you ;-)22:18
mcdoncwe need help ;-)22:18
mcdoncmy definition of a KGS is this: the "best" distribution of every distribution in the transitive set of dependencies of some set of "top-level" distributions22:20
mcdonc(where those top-level ones might be, or others that require "the world")22:21
mcdoncso in a "truly KGS", each 'package' (in the catalog index sense) would have zero or one distributions shown on its index page22:21
srichteryep, that we achieve by nailing versions22:22
mcdoncand no distributions that weren't in the transitive set of dependencies of the top-level thing would exist in the index pages of the KGS22:22
*** darrylcousins has joined #zope3-dev22:22
mcdoncs/top-level thing/top-level things22:22
srichteryeah, but that only works, if you can overlay indices22:23
mcdoncsrichter: no, it works if you don't too, you just need to be willing to tell people to use another index for things that aren't in the KGS22:23
mcdoncbut we've been over that before ;-)22:23
srichterbtw, I have real troubles to understand the meaning of "transitive set"; I looked at the disctionary for refined definitions, but I am still confused22:25
*** jsadjohnson has quit IRC22:26
mcdonci'm using it to mean the "working set" (in the pkg_resources sense) of an otherwise clean Python installation that has had every "top-level" package installed into it.22:28
mcdoncthe result of instalinng all of the right versions of every root of the working set22:28
mcdoncand no more22:29
mcdoncs/every root of the working set/every "top-level" package22:29
*** rcrafton has joined #zope3-dev22:29
mcdoncwhere the set of "top-level packages" is defined by us22:29
mcdonciow "we promise that zope.pagetemplate and all of its dependencies are installable from this index"22:30
mcdoncand we promise that if you try the same thing tomorrow, you'll get the same set of eggs22:30
*** jsadjohnson has joined #zope3-dev22:30
*** agroszer_ has quit IRC22:34
*** RaFromBRC has quit IRC22:46
*** schwendinger_ has joined #zope3-dev22:49
*** cursor has quit IRC23:04
*** schwendinger has quit IRC23:05
*** RaFromBRC has joined #zope3-dev23:06
*** darrylcousins has quit IRC23:08
*** darrylcousins has joined #zope3-dev23:10
*** regebro|food is now known as regebro23:17
*** stub has quit IRC23:18
*** MJ has joined #zope3-dev23:18
*** alecm has quit IRC23:19
*** spython has joined #zope3-dev23:22
*** natea_ has joined #zope3-dev23:24
*** natea|boston has quit IRC23:24
*** mcdonc has quit IRC23:24
*** pelle__ has quit IRC23:29
*** mcdonc has joined #zope3-dev23:33
mcdoncsrichter: i've made a distribution for zope.tal 3.4.1 and uploaded it to the cheese shop.  I also made the tag in svn (which is what I released from).23:34
mcdoncfuck.  the description and changelog isnt present on the pypi pages.23:36
mcdoncaha.  'python register' did the trick.23:39
*** Jell-O-Fishi has joined #zope3-dev23:55
*** jodok has quit IRC23:58

Generated by 2.15.1 by Marius Gedminas - find it at!