IRC log of #zope3-dev for Friday, 2005-09-16

*** SureshZ has quit IRC00:06
*** MJ has quit IRC00:08
*** MJ has joined #zope3-dev00:09
*** tarek_ has quit IRC00:12
*** clueck has quit IRC00:14
*** tarek_ has joined #zope3-dev00:16
*** srichter has joined #zope3-dev00:17
*** ksmith has joined #zope3-dev00:19
*** vlado_ has quit IRC00:19
*** ChanServ sets mode: +o srichter00:22
*** bskahan has quit IRC00:23
*** tarek_ has quit IRC00:28
*** tarek has joined #zope3-dev00:29
*** povbot` has joined #zope3-dev00:38
*** Aiste has quit IRC00:38
*** Aiste has joined #zope3-dev00:39
*** mnemoc has left #zope3-dev00:43
*** povbot has quit IRC00:52
*** J1m has quit IRC00:57
yotasrichter: hein Stephan :) Twisted integration in Zope 3 is complete ?01:16
srichternot in the trunk01:17
srichtermkerrin still needs to fix up the FTP support01:18
*** niemeyer has quit IRC01:18
yotamust I learn twisted ?01:24
yotabecause a book is out and I asked myself if I buy it01:24
srichterno, you don't have to learn twisted01:31
srichterthe default installation will provide an extended zope.conf file that let's you setup the servers01:31
srichterhowever, if you want to make use dsome of twisted's features, you need to learn twsited of course01:31
*** ksmith has left #zope3-dev01:38
yotasrichter: do you think it's possible to implement new protocol with twisted/zope3 ?01:41
srichterthis will be one of the major advantages of the merge01:41
yotafuck... graat!01:41
yotazope3 will be a devil :]01:42
*** hazmat has quit IRC01:50
*** Alef has quit IRC01:54
*** jinty has joined #zope3-dev02:07
*** yota has quit IRC02:29
*** kaczordek has joined #zope3-dev02:59
*** stub has joined #zope3-dev03:05
*** gintas has quit IRC03:47
*** _projekt01 has joined #zope3-dev04:19
*** projekt01 has quit IRC04:19
*** jinty has quit IRC04:28
*** SureshZ has joined #zope3-dev04:39
*** bskahan has joined #zope3-dev04:43
*** _projekt01 has quit IRC04:59
*** roym` has quit IRC05:04
*** tarek has quit IRC05:12
*** SureshZ has left #zope3-dev05:28
*** bskahan has quit IRC05:34
*** genconc has quit IRC07:59
*** d2m has quit IRC08:00
*** tvon|x31 has joined #zope3-dev08:13
*** tvon has quit IRC08:13
*** sashav has quit IRC08:27
*** genconc has joined #zope3-dev09:16
*** BjornT has quit IRC09:28
*** tanghus has quit IRC09:37
*** mexiKON has joined #zope3-dev09:48
*** sashav has joined #zope3-dev09:49
*** sashav_ has joined #zope3-dev09:56
*** philiKON has quit IRC10:05
*** sashav has quit IRC10:07
*** yota has joined #zope3-dev10:10
*** MJ has quit IRC10:15
*** tvon|x31 is now known as tvon10:16
*** BjornT has joined #zope3-dev10:19
*** tanghus has joined #zope3-dev10:38
*** tvon has quit IRC10:38
*** ignas has quit IRC10:42
*** MJ has joined #zope3-dev10:52
*** projekt01 has joined #zope3-dev10:53
*** ignas has joined #zope3-dev11:08
*** niemeyer has joined #zope3-dev11:12
projekt01srichter, ayt?11:18
*** seb__ has joined #zope3-dev11:25
*** seb__ is now known as yotaff11:25
*** tarek has joined #zope3-dev11:41
*** lucia12345 has quit IRC12:02
*** projekt01 has quit IRC12:02
*** clueck has joined #zope3-dev12:10
*** regebro has joined #zope3-dev12:21
*** mgedmin has joined #zope3-dev12:27
*** vinsci has quit IRC12:37
*** ignas has quit IRC12:43
*** ignas has joined #zope3-dev12:43
*** vlado has joined #zope3-dev12:44
*** vinsci|2 has joined #zope3-dev13:00
*** vinsci|2 is now known as vinsci13:04
*** clueck has quit IRC13:12
*** projekt01 has joined #zope3-dev13:28
projekt01srichter, ayt?13:52
srichterin a bit13:53
srichterI have to boot first13:53
projekt01I found a bug in the publisher's new implementation, can you help?13:53
srichteryeah, I will in a bit13:53
projekt01There is a ascii to unicode convert problem13:53
projekt01body = ('%s\n<base href="%s" />\n%s' %13:53
projekt01                                (body[:index], self._base, body[index:]))13:53
projekt01...take your boot time and tell me if you are ready ;-)13:54
*** vlado has quit IRC13:54
*** regebro has quit IRC13:55
mgedminprojekt01, and what is the problem?14:01
mgedmin'ascii_string %s' % (unicode_string,) returns a unicode string14:02
*** Alef has joined #zope3-dev14:02
mexiKONoh geez, are we still inserting that <base /> junk???14:02
mexiKONargh, can't we get rid of that in the pbulisher and make it part of the skin, if it is all that necessary?14:02
SteveAmaybe it would help to get rid of default views14:03
mgedminmexiKON, you mean you want the standard page template to have <base tal:attributes="href ..." /> somewhere in it?14:03
mexiKONwhy not?14:04
* SteveA says random stuff14:04
mexiKONi find it distrubing that the publisher mangles  my HTML14:04
SteveAyou don't need the mangling if you never use default views14:04
SteveAand always make sure that your URLs don't end in '/'14:05
SteveAthe publisher could do that14:05
*** andrew_m has left #zope3-dev14:05
mexiKONwhy not default views?14:05
mgedminmexiKON, relative links in /foo/container vs /foo/container/ vs /foo/container/index.html14:06
SteveAhttp://.../foo actually has http://.../foo/default-view  published14:06
SteveAi suppose defaullt views would be okay14:06
SteveAif you always redirect from foo to foo/14:06
*** stub has quit IRC14:06
mexiKONif self._base (see code snippet above) can be computed in the publisher, it can alsob e computed in a view class and be used by the template14:07
mexiKONi'm just arguing that the place for <base /> right now isn't ideal14:07
mexiKONSteveA, yup14:07
SteveAmake the default policy redirecting to ..../14:07
mexiKONSteveA, i was just about to suggest redirection14:07
SteveArather than using base14:07
SteveAmake an interface for views14:07
SteveAand if the publisher publishes that, it doesn't redirect14:07
SteveAor IAmAViewThatWantsMyBaseSet14:07
SteveAfor the publisher to do what it does now14:08
SteveAthere -- a 30 line proposal in the works14:08
SteveAespecially now that jim's looking at making views that want to be published provide an explicit interface14:08
SteveArather than the current vague interface14:08
mexiKONwe already have that14:08
mexiKONIBrowserPUblisher for example14:08
SteveAthat is not an interface views provide to say "i am a publishable view"14:09
SteveAthe interface is vague14:09
mgedminI do not like IBrowserPublisher14:09
SteveAit is "__call__ probably, and assume the result is unicode except when it isn't"14:09
mexiKONi still think that zope.publisher sin't the right place to set <base />14:09
mexiKONeven if we introduce a knob14:09
mexiKONwhatever that knob might be14:09
mexiKONzope.publisher shouldn't know about html at all14:09
SteveAso if a view provides not IAmAPublishableView14:10
SteveAbut IAmAViewThatWantsBaseSet14:10
SteveAthen the publisher adapts it to IAmAPublishableView14:10
SteveAthen the adapter can set base14:10
mexiKONdo we really want post-processing mangling? IIRC the cmf, plone, etc. have a <base /> in their skin templates14:11
SteveAi think this conversation would be a good thing to capture in the form of a braindump proposal14:12
* mgedmin idly wonders if backwards compatibility is important here or not14:12
SteveAbtw, launchpad now has facilities to track the metadata of specification for projects.14:12
mexiKONmgedmin, that's a good question, but i think it can easily be tackled14:12
SteveAmgedmin: not if it is to go into Zope 4 ;-)14:12
srichterwhen I discussed the base thing with Jim, he just said: .... but it is sooo useful.14:13
SteveAreally though, the publisher is something that should be re-done14:13
mgedminZope 300014:13
mexiKONsrichter, there are a lot of other things that are sooo useful14:13
mexiKONstill we don't do them for good reasons14:13
SteveAit is a component, and we shouldn't be shy of reimplementing components14:13
SteveAthe problem at present is, the interface of published stuff is too vague14:14
SteveAi'm +1 on making <base> more explicit14:14
SteveAhow that can be done is something that needs wider discussion14:14
SteveAthe best way would be a short proposal, and a ping to the list after that14:14
srichterlook, I am just giving you his opinion, nothing more14:15
* mexiKON wants generation of html where it belongs: in templates14:15
mexiKONsrichter, yeah, no problem14:15
* SteveA puts a $10 bounty on a proposal for this14:16
mexiKONsrichter, btw, already voted? ;)14:16
mexiKONmgedmin, backward compat is pretty easy. right now, the publisher only inserts base if there isnt' a base in there already, so we update templates to generate the base, generate deprecation warnings in the publisher and remove it from the publisher in 2 releases14:18
SteveAi'd say... make publisher generate base only for views that don't have an explicit "I'm a publishable view" interface14:19
SteveAso ones that have that new thing never get <base>14:19
mexiKONi'm not sure if that knob is appropriate14:20
mexiKONa browser page from a template is could say "I'm a publishable view" and its skin macro still doesn't happen to provide <base />14:20
mexiKONin fact, some customizer comes in, supplies its own skin that doesn have <base />. then the browser page makes a false statement14:21
mexiKONok, lemme rewind14:22
SteveAwe're talking deprecation here.  and new stuff shouldn't use deprecated features.  IAmAPublishableView is new stuff.14:22
mexiKONbut IAmAPublishableView has little to do with the skin macro it uses14:22
mexiKONa browser page could provide IAmAPublishableView and still use a skin (because that's a deployment thing, a skin) that doesn't have <base />14:23
mexiKONplus, it seems to me a bit overengineered14:23
mexiKONwe just say that in the future, skin templates are required to handle <base /> themselves14:23
mexiKONfor a period of 2 releases, we still support older skin templates that don't have it14:24
projekt01I think it's a good idea that the page template is responsible for the <base /> tag.14:25
projekt01But how can I fix this encoding error right now? Any idea if there is more like this invoked?14:25
mexiKONprojekt01, can you post the error somewhere?14:25
mgedminprojekt01, you did not answer my question14:25
projekt01mgedmin, sorry I didn't see it14:26
projekt01mexiKON, I try to reproduce it, it was on a customer server.14:26
mexiKON*shrug* runniing the trunk on a customer server?14:27
projekt01mgedmin, ok, that's a good question, I try to setup it again...14:27
projekt01mexiKON, there is a test server because they use i818n with french.14:28
mgedminprojekt01, well, my question was "what is your problem"14:29
mgedmin'cause that line of code you quoted looks right to me14:29
mgedmin(unless an 8-bit string appears somewhere)14:30
projekt01mgedmin, yes I setup the app on my workspace, will take a couple minutes, I'm right back with the error message.14:30
projekt01mgedmin, there is a asscii encoding error14:30
*** jinty has joined #zope3-dev14:33
*** stub has joined #zope3-dev14:33
SteveAis self._base unicode?14:34
mexiKONself._base is request.getURL()14:37
SteveAso, that'll be unicode i think14:38
SteveAhence the encoding error14:38
SteveAascii should not be miscable with unicode.14:38
SteveAthat has been a problem for python since unicode was introduced14:38
mgedminSteveA, you get encoding errors if and only if you mix two strings with non-ASCII characters -- one unicode string, one 8-bit string14:38
SteveAa solution for zope would be to insist that no non-unicode comes back from views14:39
SteveAnow, the IAmAPublishableView interface would help here14:39
mgedminpure 7-bit ascii strings are completely interchangeable with unicode strings that only have ascii chars14:39
SteveAas the interface would no longer be vague14:39
mgedmineverywhere (except zope.schema)14:39
mgedminSteveA, what about serving binary files?14:39
SteveAalso, using a bytesdata() type explicitly for non-unicode stuff14:39
SteveAwould help a lot14:39
mgedminrestricting views to unicode is not a solution14:39
SteveAi'm not saying that14:39
SteveAi'm saying that the interface should be clear and specific14:40
SteveAperhaps the interface would say "always return something along with its encoding"14:40
SteveAor it would say "IAmAPublishableUnicodeProvidingView"14:40
SteveAand there would be another to provide bytes14:40
SteveAthe bytes / unicode thing has caught launchpad folk out before now14:40
SteveAfor example, generating RDF stuff, setting the content type to application/xml-rfd14:41
mgedmindo you think that restricting a single view to only one or the other is reasonable?14:41
SteveAand the publisher expects a string, not unicode14:41
SteveAso, when you return unicode, it barfs a lot14:41
*** Alef has quit IRC14:41
SteveAyes, a view should return either unicode or bytesdata14:41
* mgedmin agrees that views like /get?doc=file.txt, /get?doc=file.png are ugly14:42
SteveAthat view needs to take responsibility for saying what charset / encoding it will use14:42
SteveAso, no problem there14:42
SteveAit just needs to take responsibility14:42
* mexiKON wonders why the str type doesn't suffice for bytesdata14:42
SteveAor, have an API to tell the publisher when the publisher should take responsibility and when it should not14:42
mgedminmexiKON, the problem is that sometimes you use str for pure ASCII data, and sometimes you use str for bytesdata14:43
SteveAmexiKON: because people do not consistently use unicode for "human readable text strings"14:43
mgedminthere ought to be two distinct types14:43
mexiKONthere are. unicode and str14:43
mgedminit is OK to mix Unicode and str-that-is-absolutely-guaranteed-to-only-contain-ascii-data14:43
SteveAmexiKON: python basically encourages you to use ascii strings for human readable ascii text14:43
SteveApython is flawed here14:43
mgedminit is a BUG if you mix Unicode and str-that-might-sometimes-contain-nonascii-data14:43
mexiKONok, so it should be discouraging14:43
mexiKONmgedmin, right14:43
SteveAso, zope should work around it by providing a bytesdata type14:43
SteveAso people can use unicode OR bytesdata14:44
SteveAand plain str becomes deprecated, for publishing14:44
mexiKONIMO, mixing str and unicode should never have been allowed. such things as sys.defaultencoding etc. are just plain evil14:44
projekt01mgedmin, on my machine is everything working ???14:44
mgedminprojekt01, do you have a traceback somewhere in a logfile perhaps?14:46
mgedminmexiKON, +1 for sys.defaultencoding14:46
mgedminmexiKON, +1 for sys.defaultencoding being EEEEVIL14:46
projekt01mgedmin, I will check the z3 trunk verison14:46
projekt01mgedmin, Ok I go to get it.14:46
mgedminmexiKON, -1 for requiring me to sprinkle my source code with u'...' when all I need is a pure ASCII string constant14:46
wiggycan't you tell python everything in your script is unicode?14:47
mexiKONbut notice that that ASCII string constant is really bytes... like in C, a chain of characters14:47
mgedminthat's an implementation detail I do not care about14:47
mgedminfor all I care it could pack 8 7-bit characters into 7 bytes of memory14:48
mexiKONwell, it's not14:48
mexiKONyou want an ascii constant14:48
mexiKONyou make it either unicode or bytesdata (==str)14:48
SteveAbytesdata derives from str14:48
* mgedmin wishes for an asciistr type that raises exceptions when you try to get 8-bit characters into it14:48
mexiKONright, that is feasible14:49
mexiKONhave asciistr, str, unicode14:49
mgedminis basestring an abstract class?14:49
mexiKONi think so14:50
mexiKON>>> basestring('')14:51
mexiKONTraceback (most recent call last):14:51
mexiKONTypeError: The basestring type cannot be instantiated14:51
mgedminthen maybe we could have an asciistring type that is a subclass of basestring and a superclass of both str and unicode14:51
mgedminit can be automatically promoted to either of them14:52
SteveAhaving asciistr isn't useful i think14:52
mgedminsadly, forbidding any combination of str and unicode would break backwards compatibility14:52
SteveAbecause 'foo' is still a str14:52
mgedminunless from __future__ import pure_strings?14:52
* mgedmin is way out of his depth and will shut up now14:53
SteveAlunchtime, lithuanians14:53
mexiKONmgedmin, +1 for pure_strings14:54
mexiKONfoo = ''14:54
mexiKONtype(foo) == asciistr14:54
mexiKONfoo = b''14:54
mexiKONtype(foo) == str == bytesdata14:54
* mexiKON goes to lunch14:55
*** andrew_m has joined #zope3-dev14:59
projekt01mgedmin, I cant' reproduce it, on the customer test server is everything OK after restart.15:01
projekt01mgedmin, Is it possible that a runtime error happens and required a restart? The view which runs into the error is working well.15:03
*** benji_york has joined #zope3-dev15:15
*** tvon has joined #zope3-dev16:03
*** stub has quit IRC16:07
*** mexiKON is now known as philiKON16:10
projekt01benji_york, do you know the zpkg well?16:37
benji_yorknope, I need to learn it, but haven't yet16:39
projekt01perhaps you can answer a question. Is it possible to use zpkg for export a workspace described in a source map and use the exported tar extract for building another installer?16:41
philiKONprojekt01, the tar already contains an installer16:42
benji_yorkhmm, I really don't know16:42
projekt01Yes, I know, but I like to use another one.16:42
projekt01another/better ;-)16:42
philiKONzpkg's tar archives are really designed to be used with the zpkg installer16:43
*** anguenot has joined #zope3-dev16:44
projekt01but the zpkg installer isn't good a installer since it will install everything to the python install dir and you can't have different version16:44
projekt01It's more a UI for a install script ;-)16:45
*** fdrake has joined #zope3-dev16:45
philiKONprojekt01, that's not true16:46
philiKONprojekt01, by *default*, IOW, if you give no other arguments, it installs into python site-packages16:46
philiKONprojekt01, but you can specify --home16:46
philiKONe.g. python --home=/usr/local/Zope-3.1.016:46
philiKONprojekt01, in fact, the generated Makefile (it's generated by the configure script) does exactly that16:47
projekt01philiKON, I can't install z3 3.1 and 3.2 for python 2.4 with the installer!16:47
philiKONsure you can16:47
philiKONtry the line i gave you above16:47
projekt01both of then install the source to the python 2.4 location16:47
projekt01Ah, Ok, I have to try this16:48
philiKONthe --home switch changes the installation target16:48
fdrakeprojekt01, what's the right way to tell Subversion to run in an alternate locale?  I tried "LANG=de svn help", but that failed (it told me it couldn't set the locale).16:49
fdrakeit might be I just don't have the alternate locales installed on my Ubuntu box, also16:50
*** deo has joined #zope3-dev16:50
projekt01fdrake, I don't know, I'm also not sure why on my windows some messages are in german and some in english, reported form the command "svn cat path"16:50
philiKONfdrake, afaik LANG is the right env var16:50
philiKONfdrake, you probably indeed need to install svn locales16:51
fdrakethe language mix is likely incomplete i18n/l10n in svn16:51
projekt01fdrake, I think so16:51
projekt01fdrake, did you see my other question about using the zpkg in relation to another install builder?16:53
* fdrake is trying to figure out what he needs to install...16:53
projekt01This bugfix is working for me:16:54
projekt01if "directory" in err or "Verzeichnis" in err:16:54
fdrakethat won't help anyone from Nuxeo, though, would it?  ;-)16:55
projekt01No, let add them the french word for directory ;-)16:55
philiKONfdrake, maybe zpkg should set the LANG env var to 'C' for svn16:56
projekt01Really bad that no code get reported there additional to the message.16:56
projekt01Is there no other way to find the difference between directory and files?16:57
fdrakethat's likely more reliable than me figuring out how to set locales properly, and get translations installed16:57
fdrake"svn info" only works on checked-out stuff16:58
philiKONfdrake, btw, ever thought about using the subversion python bindings?16:58
fdrakeand that code is trying to figure out what to do with what's going to be checked out16:59
fdrakei hadn't, since they might not be installed16:59
fdrakealso, Jim's made them segfault a bit while playing with them lately16:59
fdrakethey're built using SWIG, which I consider reason enough to avoid them16:59
projekt01I think a lang env would be OK for now.16:59
fdrakecan you try running zpkg w/ LANG=C for now, and let me know if it fixes that problem for you?17:00
philiKONfdrake, yeah17:00
projekt01philiKON, perhaps python bindings are a good idea for this.17:00
projekt01Lang=C, you mean LANG=DE17:01
fdrakeno, LANG=C, to avoid the localization17:01
fdrakei'm searching for a string in the output17:01
fdrakethat string gets translated if I do nothing17:01
projekt01Ah, Ok let me setup and try it, I will come back to you later17:01
philiKONfdrake, btw:
philiKONso, LANG=de isn't enough17:02
philiKONneeds to be de_DE17:02
*** sashav_ has quit IRC17:02
fdrakephiliKON, LANG=de_DE doesn't work for me, so I'm probably missing some bit of locale support17:08
fdrakeLANG=C works fine though  :-)17:08
projekt01fdrake, can you tell me the command where you are suing with LANG=C ?17:13
fdrakeLANG=C svn help17:15
fdrake(LANG=C should use the untranslated text straight from the executable, so it should work for everyone)17:15
philiKONprojekt01, are you on windows?17:15
mgedminfdrake, you might want to use LC_ALL=C17:16
mgedminas far as I understand POSIX, LC_MESSAGES overrides LANG, while LC_ALL overrides both LC_MESSAGES and LANG17:16
philiKONprojekt01, set LANG=C17:16
fdrakeah, even better17:16
mgedmin(and then there's GNU's non-POSIX $LANGUAGE which overrides everything...)17:16
projekt01Ok set LANG=C is working17:17
projekt01svn: URL 'svn://' refers to a directory17:17
projekt01is the reported message17:17
projekt01no "Verzeichnis" anymore17:18
philiKONnot having used the zope 3 catalog, how does it do less than zope 2's zcatalog as it is always said?17:18
*** kaczordek has quit IRC17:19
*** deo has quit IRC17:19
projekt01fdrake, did you see my question about using the exported build source for another installer?17:19
fdrakei see the emails, but haven't read them all yet17:20
fdrakeone thing at a time  :-)17:20
*** regebro has joined #zope3-dev17:22
projekt01no problem17:22
fdrakeone down, and the kids only got into one fight...17:30
andrew_mcan it be that something with ftp is broken in zope svn? something with "__call()__ takes 3 arguments.. " when trying to connect..17:31
srichterprojekt01: are you getting my private messages to you?17:33
projekt01yes, I'm reading the source....17:34
fdrakeprojekt01, have you looked at the -t command line option for zpkg?17:35
projekt01no, what's that?17:35
fdrakeit builds the tree w/out making a tarball17:36
fdrakewould that help with what you're doing?17:36
fdrakeyour message wasn't very clear; sorry.17:36
projekt01Ok, and how can I build a distribution form that source? with build_ext?17:37
*** ignas has quit IRC17:37
fdrakeThis is where I got confused; I'm not sure what you're trying to do.17:37
*** ignas has joined #zope3-dev17:37
fdrakeThat tree is essentially an unpacked source distribution as it is.17:38
projekt01I like to use the zpkg for asamble a source from different repositories and this exports use to build a running version17:38
fdrakeEverything that distutils can do should already be supported.17:38
projekt01It looks very different with all this own folders for sub-packages.17:38
fdrakeyou want to run directly from the zpkg-constructed tree?17:38
fdrakeI'm afraid that was never a use-case for zpkg here.17:39
projekt01Yes, I see17:39
fdrakeYour best bet is to install from that tree, or build a platform package.17:40
fdrakebdist_dumb, or bdist_wininst17:40
projekt01Then zpkg doesn't export the code in a usable way for build and running it out of the exported structure17:41
fdrakeyou can build using "python build"17:41
fdrakebut that just compiles things and copies the package tree intro build/*17:41
projekt01Yes, that's what I'm looking for.17:42
fdrakebut that doesn't look like an installation either17:42
fdrake(but closer)17:42
fdrakeyou can use "python install --home <target>" to install outside the Python installation17:43
fdrakemight require Python 2.4 on Windows17:43
projekt01I just need a working/running build17:43
fdrakeI *think* that's when I added that.17:44
projekt01Ok, I use Python 2.417:44
fdrakewe're going to be looking at creating a "build-out" mechanism that uses zpkg to do some of the work17:45
fdrakebut I don't know just when17:45
fdrakehopefully Benji and I will have time to talk about it next week to figure out how we want it to work17:46
projekt01Is this not correct, zpkg is exporting source code from different repositories defined in map's to one workspace. In a next step you can build a distribution.17:46
fdrakeGuess you missed my zpkg presentation at Wednesday's Fredericksburg ZPUG.  :-)17:46
fdrakeThat is not correct.17:46
projekt01of corse ;-(17:46
andrew_merror: uncaptured python exception, closing channel <zope.server.ftp.server.FTPServerChannel connected at 0x408787cc>(exceptions.TypeError:__call__() takes exactly 3 arguments (4 given) [/home/andrew/opt/ZopeX3-svn/src/zope/server/linereceiver/|service|43] [/home/andrew/opt/ZopeX3-svn/src/zope/server/ftp/|cmd_cwd|175] [/home/andrew/opt/ZopeX3-svn/src/zope/server/ftp/|type|45] [/home/andrew/opt/17:47
fdrakezpkg exports lots of different packages from various repositories (including the local disk) into separate workspaces (for zpkg purposes), and then knits material from those into a distribution17:47
andrew_many ideas anyone?17:47
projekt01fdrake, the first part is Ok, but is there no good way to use this workspace/export as a base for other install builder tools? that's what I'm looking for.17:49
srichterandrew_m: yep, it's a failure due to the publisher refactorings17:49
fdrakeI don't think so.17:49
srichterandrew_m: somewhere an outstream is specified where it is not expected17:50
fdrakeMost of the time you can point zpkg to existing checkouts, which can be shared with a working environment for development or distribution building.17:50
andrew_msrichter: aha.. and can you think of a quick and dirty fix? this happens with a fresh svn checkout of zope / fresh instance when connecting via ftp17:50
fdrakebut that doesn't need zpkg17:50
srichterandrew_m: find the spot where it happens, don't pass in the outstream and check in the fix with a test ;-)17:51
projekt01fdrake, I'm interested in the dependency where get recognized by the zpkg and will get automaticly exported.17:51
andrew_msrichter: sounds neither quick nor dirty17:52
fdrakeprojekt01, look in zpkgtools/ in build_distribution()17:54
srichterandrew_m: it will be the only way17:54
srichterandrew_m: you should not be afraid of fixing bugs17:54
andrew_msrichter: any idea if that problem is also there in 3.1 RC2?17:55
*** ignas has quit IRC17:56
srichterandrew_m: definitely not; this error is due to recent publisher changes17:57
andrew_msrichter: ok, thanks.. we'll revert to RC2 here then for now17:58
projekt01fdrake, Ok, when is this method used and what's the result?17:58
projekt01srichter, is there a named sample or did I overlook it?18:00
*** d2m has joined #zope3-dev18:00
srichterandrew_m: you should at least report the problem in the collector18:00
srichterotherwise 3.2 will be broken for you too18:00
srichterprojekt01: I am still finishing the named code18:01
andrew_mi will18:01
projekt01srichter, ah Ok, it looks very clean right now18:01
srichterprojekt01: you mean the "pagelet" namespace, right?18:01
srichterI have this ready in a few moments18:02
projekt01there is only pagelets: used in the sample. right?18:02
fdrakeprojekt01: that's the main function that builds the output tree; it's called from the run() method18:02
projekt01fdrake, Ok, but not runnable in some way18:02
srichterprojekt01: yep18:04
projekt01drake, this means the zpkg isn't usable for doing only the first step (assamble code from repostories defined in maps) for other install builders.18:04
fdrakeI'm not surprised.  The tree zpkg puts together is a bit on the strange side from where I stand.  Jim and I have spent a lot of time hashing this out, and I'm not sure that either of us are really happy with it (though I suspect Jim is happier with it than I am).18:05
projekt01fdrake, because the logic of this exported structure is only known in additional python code (means only zpkg knows how to build a ditribution)18:06
fdrakeI'd be happy to see suggestions for improvements.  I'd like to encourage you to write up a preferred tree structure and post to the zpkg mailing list about it.18:07
projekt01I don't care about the tree structure. For me it is much more important what I can do with this structure with less then possible coding.18:08
fdrakeThen what you want is a much simpler structure, right?18:09
projekt01I like to see that the very usefull collection part of zpkg is useable for other tasks.18:09
projekt01Why do we not use the same structure like the repositories have?18:10
projekt01This source is running out of the box after build the C-extensions18:11
fdrakeJim wanted to keep each component separately installable, so each one actually includes a with everything it needs.18:11
fdrakeIf we drop that requirement, then we can have a tree that's much simpler and more usable for other tasks.18:11
projekt01Ah, that's why we get a very proprietary structure.18:12
fdrakeThe distribution itself is unusable for anything but installation as a result.18:13
projekt01I'm not sure if it's possible to use a simply structure, but if so, we can use zpkg for other tasks too.18:13
fdrakeI consider this unfortunate myself, but that's personal opinion.18:13
projekt01I really, really like to see zpkg working for running checkouts where dependency get correct recognized ;-)18:14
projekt01The other (build dist) part should be decoupled in some way18:14
fdrakeWe probably could use a simpler structure, but we'd have to handle metadata differently in the distribution than in the checkouts (at least somewhat).18:14
projekt01Then the SETUP.cfg and DEPENDENCY.cfg make more sense to me.18:14
fdrakecurrently, SETUP.cfg is only needed at build/install time; it's used to drive distutils.18:15
fdrakeDEPENDENCIES.cfg is only used for distribution construction, but defines constraints on future versions of the implementation18:16
fdrakePACKAGE.cfg is strictly a distribution construction thing, and can (and should!) be avoided for most things18:17
projekt01Does zpkg provide a "install additional packages" for a installed z3 concept?18:17
fdrakePUBLICATION.cfg is similar18:17
fdrakenot sure what you mean18:17
fdrakeas in "install more stuff into my Z3 installation", or "install more stuff into my Z3 instance"?18:18
projekt01Can you use zpkg for checkout additional z3 packages like the ldap package at svnzope.org18:18
projekt01"install more stuff into my Z3 instance"18:19
fdrakeif it's in CVS or SVN, it should be able to use it18:19
fdrakethat depends on what you want done with ZCML slugs18:19
projekt01that's what admins will do ;-)18:19
fdrakecurrently, ZCML slugs get installed into zopeskel/etc/package-includes/, which only makes sense for the installation18:20
fdrakeif the SETUP.cfg for a package says to use etc/package-includes/ instead, it should be installed into the instance18:21
fdrakeI think it's a problem that you have to make that decision in the package, but that's the current state18:21
projekt01you mean if you checkout additional packages with zpkg to a instance?18:21
projekt01Yup, I agree, this should be a checkout descision18:21
fdrakeno; this is an installation thing; you don't checkout packages into instances18:21
fdrakeeither way, it's installation of the package18:22
projekt01sorry, I mean checkout a package with zpkg into a instance. (install)18:22
fdrakethe current pattern should be:18:22
fdrake1. install Z3 from the distro18:22
fdrake2. install any additional packages you want to share among instances18:23
fdrake3. create instances18:23
fdrake4. install instance-specific packages18:23
fdrakeyou control where packages get installed (in all cases) using "python install --home <where>"18:24
fdrakebut which installations will make sense depends on how ZCML slugs are destined based on SETUP.cfg18:24
projekt01you say in step 4.) you can install additional packages to a specific instance, right?18:25
fdrakeI have some ideas on how to improve that, but it means the zpkg "runtime" portion remains involved in installation, or something gets added in Z3 as a helper for package installation18:25
fdrakethose packages need to specify etc/package-includes/ as the destination for ZCML slugs, though18:26
projekt01Then I can install the z3 trunk itself to a location without the build step, right?18:26
fdrakeyou can use "python install --home <instance>" to add packages to an instance18:26
projekt01It's more a zpkg and SETUP.cfg controlled checkout.18:27
fdrakethe build step is implied by the install step for distutils18:27
fdrakelike Z3 itself?18:28
projekt01Yes, but do the dependency get recognized with install --home. I don't think so?18:28
fdrakedependencies are handled when the distribution is built, and only then18:28
projekt01Do you know the active state perl installer?18:28
projekt01perhaps there is anpther name for this package tool or whatever this is called.18:29
*** J1m has joined #zope3-dev18:29
projekt01Ok, I try the install --home option and try to use this source for use a different install builder.18:31
fdrakethat's probably your best bet for now18:32
*** sashav has joined #zope3-dev18:34
projekt01Hope to see zpkg produce reusable workspace source in the future ;-)18:34
projekt01fdrake, many thanks18:34
projekt01fdrake, or perhaps we have to start a that tool we spoke about with benji_york for checkout project based source code from repositories.18:36
projekt01I guess a project based source code checkout and build tool would be very useful18:38
*** GaryPoster has joined #zope3-dev18:39
fdrakeyes, it would be18:40
J1msrichter, ayt?18:40
fdrakein fact, we have such a thing internally18:40
fdrakestill might not be what you're looking for, though  :-)18:40
*** GaryPoster has quit IRC18:41
*** GaryPoster has joined #zope3-dev18:43
benji_yorkfdrake and projekt01, it's not internal any more, the buildout was put up on the sandbox a couple days ago, I intend do announce its availability today18:43
srichterJ1m: yep18:43
J1msrichter, what are the plans for 3.1?18:45
srichteroh, darn, I totally forgot18:46
fdrakethat's a good plan  :-)18:46
projekt01benji_york, cool, does buildbot support additional repositories?18:47
srichterI was going to ask again if anyone needed to make more chackins for RC3?18:47
fdrakesrichter: not me18:47
srichterGaryPoster: benji_york: fdrake: Are you guys all set?18:47
GaryPostersrichter: yes, I'm actually really eager for it :-)18:47
srichterI remember there was a range of bugs that needed addressing18:47
GaryPosterAll of the ones I was following (and that we intended to address for this version) are handled18:48
srichtergreat; I think you mainly worked out the locale issue right?18:48
benji_yorksrichter, I am18:48
benji_yorkprojekt01, I'm not sure what you mean by "additional repositories"; if you mean can it fetch things from more than one svn repo, then yes18:50
projekt01Yes, cool18:50
benji_yorkprojekt01, there is an example buildout at svn://
benji_yorkjust check that out, change to the directory you checked it out into and run build.py18:52
benji_yorkit will take a while because it builds Python and libxml2 (as examples of how to do things like that), but after it's done you can look around18:53
benji_yorkthere are no docs and some rough edges, but it's a good start18:53
projekt01benji_york, what does the skeleton mean in the buildbot?18:53
srichterGaryPoster: Can you give me a sentence of what you fixed?18:54
srichterfdrake: did we update zpkgtools since RC 2?18:54
benji_yorkit is used by the "instance" recipe to build an instance, other than that it is not required18:55
*** MJ has quit IRC18:56
projekt01benji_york, does it mean if I like to run/build tiks packages I have to place them also to the package-includes?18:57
GaryPostersrichter: sure.  ported Stuart Bishop's work in pytz and zope.i18n to address broken and misleading timezone code (i.e., one issue was broken and another was misleading :-) ).  Fixed apidoc to handle extended paths, so that zope packages could be installed in software instances without apidoc losing knowedge of the core zope code.18:57
benji_yorkprojekt01, if you have ZCML I wouldn't add it to the skeleton, I'd have the recipe add the slug to the package-includes of the instance (but I'm not entirely sure of what/why you're asking)18:59
*** baldtrol has joined #zope3-dev18:59
srichterGaryPoster: I think you did not merge the API doc fixes to the 3.1 branch18:59
srichterI don;t see anything in my archives about that18:59
GaryPosterHm, think I did.  will check19:00
GaryPoster(That's where I needed it :-) )19:00
GaryPosterrevision 3834819:01
srichterok, thanks19:01
fdrakesrichter, there have been changes, but I don't think you should update anything for 3.1.0 at all19:01
projekt01benji_york, why is the build bot not a z3 app and has some views for a UI and uses the twisted task scheduler for run nightly builds ;-)19:03
benji_yorkyou'll have to talk to the buildbot people about that :)19:04
srichterprojekt01: because it is a well-known twisted project; there is no home-made code19:04
projekt01Ah, Ok19:05
*** vlado_ has joined #zope3-dev19:14
*** MJ has joined #zope3-dev19:24
*** M1 has joined #zope3-dev19:27
*** MJ has quit IRC19:28
*** M1 is now known as MJ19:28
*** christi has joined #zope3-dev19:30
*** christi has left #zope3-dev19:31
J1msrichter, what are the plans for 3.1? :)19:32
srichterJ1m: you mean now that I speeded out RC 3?19:33
srichterI hope I just make it the final next week19:33
J1mGreat. Thanks. :)19:33
srichterI am getting faster with the release thingy :-)19:37
*** tarek has quit IRC19:39
srichterthe trick with publishing content on is definitely to specify the effective date when changing the state19:41
*** yotaff has quit IRC19:44
*** vlado_ has quit IRC19:44
*** niemeyer_ has joined #zope3-dev19:45
*** niemeyer has quit IRC20:01
*** niemeyer_ has quit IRC20:05
GaryPosterMany thanks srichter!!  And yes, you generally need both dates on, AFAICT.20:48
srichterok, I always only specify the first and this seems to work recently21:02
*** anguenot has quit IRC21:06
*** GaryPoster has quit IRC21:13
*** sashav has quit IRC21:18
*** sashav has joined #zope3-dev21:19
*** GaryPoster has joined #zope3-dev21:22
*** bradb has quit IRC21:22
*** baldtrol has left #zope3-dev21:23
*** bradb has joined #zope3-dev21:23
*** efge has quit IRC21:26
d2mThere is a 'private' SWRelease file now in folder, actually blocking anonymous access to the folder (and the downloads)21:27
d2msrichter, ok - its been published that second21:29
*** clueck has joined #zope3-dev21:32
srichteryep, it was probably Tim21:32
*** regebro has quit IRC21:35
*** Theuni has quit IRC21:47
*** Alef has joined #zope3-dev21:50
fdrakeprojekt01: What version of Windows are you using?21:52
fdrake(trying to figure out the quoting problem you reported)21:53
*** jinty has quit IRC21:57
*** Aiste has quit IRC22:01
*** BjornT has quit IRC22:02
*** Alef has quit IRC22:03
*** clueck has quit IRC22:07
*** mgedmin has quit IRC22:11
philiKONsrichter, i think you have a "bug" in the release announcement for zope 3.122:15
philiKONsrichter, isn't part of zope 3.122:15
srichterok, I'll remove it next time22:16
philiKONi asked this earlier, but it was eaten by the zpkg discussion:22:24
philiKONi haven't looked too closely at the zope 3 catalog; it is always said that it does less than zope2's zcatalog. what does it do less? i mean, it still takes care of indexes and allows queries...22:25
srichtercan I do <tal:block repeat="item ..." replace="item" /> ?22:26
philiKONafaik yes22:27
philiKONthe execution order is something like  define, repeat, replace|{attributes, content}22:28
projekt01fdrake, ayt?22:36
projekt01fdrake, win XP22:37
fdrakeok, did you get my email?22:40
fdrakethere's another follow-on question there22:40
projekt01in the at line 317 I'm using  (cmd = 'svn cat %s' % url) intead of (cmd = 'svn cat "%s"' % url)22:40
fdrakedo you actually get an error when the quotes are included at that point?22:41
projekt01didn't read the mail, will do it in 15 minutes...22:41
fdrakewhat, specifically?22:41
projekt01no error just abort22:41
projekt01an there was a report that a temp folder where generated22:42
projekt01I think it is something like a general finish on error where I run into.22:43
fdrakeis this with a specific Zope 3 release?22:44
projekt01fdrake, I'm back, this was with 3.0.1b.22:50
projekt01Just like described in srichters readme for a windows build on www.zope.org22:50
fdrakeok, I'll see what turns up22:52
projekt01fdrake, I will run the code again now. just wait a minute.22:52
fdrakewould that be the MakingARelease page?22:53
projekt01Yes, there is a small link pointing to windows notes from srichter I guess22:53
srichterno, it's from Tim22:54
projekt01fdrake, the code is running now with svn cat "the svn url".22:54
projekt01I was wrong with this issue. It's working since I set the right language.22:55
projekt01fdrake, sorry for that, but now the zpkg works on windows too. with "set LANG=C" set in the DOS box22:57
srichterprojekt01: will it be okay, if I move the pageletchooser and boston skin to branch for now and check in my pagelet redesign?23:00
srichterprojekt01: I need your help in fixing them; I am quiet lost ;-)23:01
fdrakeprojekt01, that's good news; thanks!23:01
projekt01fdrake, thanks to you.23:02
* srichter feels like projekt01 is ignoring him today23:04
projekt01sorry didn't you see my private window?23:05
projekt01srichter> Roger, will it be okay, if I move the pageletchooser and boston skin to branch for now and check in my pagelet redesign?23:05
projekt01<srichter> this way you can help me fix the other components :-)23:05
projekt01<srichter> I am quiet lost, especially with the boston skin23:05
projekt01<projekt01> Ok, we can do this23:05
projekt01<projekt01> srichter are you moved this part allready? I have a lot of changes for the package. Should I checkin this first?23:05
projekt01<srichter> ping23:05
projekt01<projekt01> pong23:05
projekt01<projekt01> I guess you have to move th23:05
srichterI am not getting those messages; strange23:05
*** srichter has left #zope3-dev23:05
*** srichter has joined #zope3-dev23:06
*** ChanServ sets mode: +o srichter23:06
projekt01can I commit my pending work, that we have them in the branch too?23:06
projekt01I have still uncommited code for the boston skin and the skintools in my workspace23:06
srichterprojekt01: yeah, switch your working copy to the branch and commit there23:07
srichterprojekt01: currently the branch is a copy of the trunk, so you can do an svn switch to it23:07
projekt01Ok, did you allready add a branch?23:08
projekt01Do you know, can I switch my workspace without loosing my uncommited work?23:09
srichtersvn switch ...23:09
*** fdrake has left #zope3-dev23:11
srichtersvn switch svn+ssh:// .23:11
srichterprojekt01: do you have a moment?23:18
projekt01 Yes, just running the functional tests and I'm read to commit23:19
projekt01srichter, Done, I'm ready to work on the pagelet till the german elections starts on sunday night ;-)23:25
srichterI am about to check in the new code to the trunk23:25
srichterwe then need to merge the pagelet package to the branch23:25
srichterand all hell will break loose :-)23:26
andrew_mhmm.. any idea anyone where has gone in zope 3.1c3? or will that not be available in 3.1?23:26
projekt01Cool, I will switch back to the trunk23:26
srichterandrew_m: was it in 3.0?23:26
projekt01Hey, do you know there are many out there who use this pagelet system.23:26
andrew_mdunno, been in svn trunk for ages though23:26
projekt01Hope they don't work on the weekend ;-)23:27
philiKONandrew_m, isn't in 3.1?23:27
philiKONandrew_m, isn't in 3.1.23:27
andrew_mphiliKON: at least i cannot find it under that name23:27
andrew_msearched for *lock*23:27
philiKONsorry, last sentence counts23:27
philiKONit's a statement23:27
srichterandrew_m: that does not mean anything; there is a lot of stuff in the core that is not in the release23:27
philiKONandrew_m, it's just not in the release for 3.123:27
* andrew_m tries to find an svn release with ftp working and locking23:29
srichteruse the 3.1 branch I would assume23:29
andrew_msrichter: tnx, i'll try that first23:32
srichterprojekt01: the new pagelet code is in23:33
projekt01srichter, Ok it's landed in my workspace23:35
projekt01where can we start?23:35
srichterwe need to merge the new pagelet code to the pagelet-rework branch23:35
srichterthen I think the first task should be to fix the boston skin and skintools23:36
srichterbtw, we should flatten skintools a bit23:36
projekt01if they get merged to the branch.23:36
srichterthe browser directory is too much23:36
srichterno, only merge the package23:36
srichteralso, I really want to understand the use cases for pageletchooser23:37
srichteryour use case was full of macro talk, and that did not make sense after the refactorings at all23:37
projekt01I would like to rename the skintools to ui if possible. This would also make the rework easy.23:37
srichterI think we also want to move skintools into boston for now23:38
srichterbecause noone is using it at this point23:38
srichterother than boston23:38
projekt01but this part is generic and I use it different skins without the boston skin.23:39
projekt01I think the skin and additional shared pagelet viewlets should be in different packages23:40
projekt01Is the working?23:40
*** SureshZ has joined #zope3-dev23:43
srichterprojekt01: no, I think we can remove it23:43
srichterprojekt01: it is far too complex anyways23:43
srichterthe README.txt has now a concise example23:43
srichterif you feel strong about, we could fix it though23:44
srichterprojekt01: ok, the branch now has the trunk pagelet version23:44
projekt01Ok, I agree, perhaps we can add a better demo later which is more useful23:45
projekt01Ok, I will take a look to the branch23:45
srichterI think the README.txt will be good as a simple demo and boston will be a complex one23:46
srichterprojekt01: do you mind if I give you another call?23:48
srichterprojekt01: I need to understand pageletchooser23:48
projekt01That's no problem23:48
projekt01srichter, go to for a sample use case where I can describe.23:50
J1msrichter, I can't get to the release anonymously23:52

Generated by 2.15.1 by Marius Gedminas - find it at!