IRC log of #zope3-dev for Friday, 2006-09-08

*** hazmat has joined #zope3-dev00:03
*** ChanServ sets mode: +o hazmat00:03
*** RaFromBRC_ has joined #zope3-dev00:19
*** RaFromBRC has quit IRC00:20
*** RaFromBRC_ is now known as RaFromBRC00:21
*** batlogg has quit IRC00:28
*** RaFromBRC is now known as RaFromBRC|lunch00:29
*** dokai has quit IRC00:49
*** hazmat has quit IRC00:53
*** dokai has joined #zope3-dev00:55
*** hazmat has joined #zope3-dev00:56
*** ChanServ sets mode: +o hazmat00:56
*** RaFromBRC|lunch is now known as RaFromBRC00:57
*** hazmat has quit IRC01:05
whitanybody know where the zc.recipe.testrunner lives?01:16
benjiwhit: it's in zc.buildout01:23
whitis there a shorthand way to run just it?01:24
benjito run the testrunner?01:27
benjiyou should get a script: "bin/test"01:28
whitI'm not quite to the getting a script point01:29
whitI guess what i'm asking is how to I have buildout do that for me01:31
*** yota has quit IRC01:36
*** natea is now known as natea|away01:37
*** jinty has quit IRC01:42
benjiwhit: (sorry for the lag) I believe if you use the zc.recipe.testrunner recipe like this you'll get a bin/test script when you run bin/buildout01:44
benjiwhit: this might be a better (simpler) example:
whitthanks benji01:48
*** alecm has quit IRC01:50
*** natea|away has quit IRC01:52
*** natea has joined #zope3-dev01:53
*** natea has joined #zope3-dev01:55
*** natea is now known as natea|away02:19
rockyhmm... on an IObjectModifiedEvent is there anyway to determine what fields changed?02:19
benjinot that I know of, rocky02:20
rockyi have a chicken and egg problem :(02:20
benjiis this for debugging or for actual use?02:20
rockyactual use02:21
benjiyou could have the object keep up with it's changes and ask it (in your IObjectModifiedEvent subscriber)02:21
rockyon IObjectModifiedEvent being triggered during formlib edit ... and if fields x, y, and z were modified, then field foo should be updated based on those fields... but sometimes field foo itself might be directly modified, in which case i don't want to accidentally override it's value with x, y, and z02:21
benjioh, I don't think I'd do that with a subscriber02:22
rockyperhaps i'm going about this all wrong then02:22
benjithere /should/ be some way to do that with formlib, but I don't know enough about it to point you in the right direction02:23
*** benji has quit IRC02:27
*** niemeyer has quit IRC02:38
*** alga__ has quit IRC03:05
*** niemeyer has joined #zope3-dev03:09
*** J1m has quit IRC03:12
*** markup_ has quit IRC03:32
*** wrobel has quit IRC03:33
*** niemeyer has quit IRC03:33
*** wrobel has joined #zope3-dev03:34
philiKONrocky, yes, there is a way04:10
philiKONnotify(ObjectModifiedEvent(obj, Attributes(ISample, "field")))04:12
philiKONthat indicates that the ISample(obj).field property has changed for obj04:12
philiKONISample(obj) may be an adapter, it may be obj itself04:12
philiKONso, you'd also use ObjectModifiedEvent for things like IZopeDublinCore properties, when those change04:13
* philiKON zzZzzZZzz04:15
*** philiKON has quit IRC04:15
*** projekt01 has left #zope3-dev04:29
*** RaFromBRC has quit IRC04:56
*** natea|away has quit IRC06:14
*** natea has joined #zope3-dev06:16
*** runyaga has joined #zope3-dev06:27
* runyaga smiles06:27
*** batlogg has joined #zope3-dev06:59
*** baijum has joined #zope3-dev07:08
*** stub has joined #zope3-dev07:18
*** runyaga is now known as run|away07:44
*** batlogg has quit IRC07:46
*** batlogg has joined #zope3-dev07:58
*** yota has joined #zope3-dev08:04
*** jukart has joined #zope3-dev08:08
*** eins has joined #zope3-dev08:14
*** zagy has joined #zope3-dev08:17
*** oferw has joined #zope3-dev08:24
*** natea has quit IRC08:37
*** alecm has joined #zope3-dev08:49
*** oferw has quit IRC08:51
*** zagy has joined #zope3-dev08:54
*** BjornT has quit IRC08:59
*** d2m has joined #zope3-dev09:04
*** BjornT has joined #zope3-dev09:24
*** philiKON has joined #zope3-dev09:36
*** romanofski has joined #zope3-dev09:38
*** dobee has joined #zope3-dev09:45
*** mexiKON has joined #zope3-dev09:54
*** BjornT___ has joined #zope3-dev10:00
*** philiKON has quit IRC10:02
*** kobold has joined #zope3-dev10:04
*** hdima has joined #zope3-dev10:06
*** MJ has quit IRC10:18
*** mexiKON has quit IRC10:33
*** BjornT__2 has joined #zope3-dev10:46
*** BjornT has quit IRC10:47
*** BjornT___ has quit IRC10:48
*** BjornT__ has joined #zope3-dev10:48
*** BjornT__ has quit IRC10:50
*** BjornT has joined #zope3-dev10:50
*** alecm has quit IRC10:51
*** MJ has joined #zope3-dev11:05
*** timte has joined #zope3-dev11:28
*** romanofski has quit IRC11:31
*** romanofski has joined #zope3-dev11:32
*** philiKON has joined #zope3-dev11:33
*** ignas has joined #zope3-dev11:42
*** projekt01 has joined #zope3-dev11:59
*** faassen has joined #zope3-dev12:22
*** MJ has quit IRC12:25
floxzope.modulealias: "This package has been deprecated as of 2004/02/24 and will be removed after 12 months."12:26
*** MJ has joined #zope3-dev12:27
floxi guess i can remove dependency of on this module (for the /trunk, of  course)12:27
philiKONflox, i think the date is supposed to be 2006/02/2412:28
*** ignas has quit IRC12:28
philiKONflox, yup, just confirmed12:29
philiKONdate should be 2006/...12:29
philiKONwe can't lift the dependency on it in until it's removed12:30
floxso we keep the line <include package="zope.modulealias" file="meta.zcml" /> in "/zope/app/meta.zcml"12:30
floxi commit a change for the deprecation date, ok?12:31
philiKONyes pelase12:31
*** projekt01 is now known as projekt01_12:35
*** Aiste has quit IRC12:38
*** jinty has joined #zope3-dev12:48
*** volvox has joined #zope3-dev12:58
*** Aiste has joined #zope3-dev13:01
*** ignas has joined #zope3-dev13:06
*** volvox has quit IRC13:24
*** volvox has joined #zope3-dev13:24
*** ignas_ has joined #zope3-dev13:28
*** ignas has quit IRC13:29
*** ignas_ has quit IRC13:29
*** ignas_ has joined #zope3-dev13:30
*** dobee has quit IRC13:32
*** mkerrin has joined #zope3-dev13:49
*** jhauser has joined #zope3-dev14:04
philiKONflox, btw, you can put zcml:condition on the tag itself14:06
philiKON<browser:layer zcml:condition />14:06
philiKONno need to wrap it in a configure element14:06
philiKONflox, ayt?14:13
*** ofer has joined #zope3-dev14:14
floxyes, i am here14:16
floxi was looking for visual studio :-~14:16
floxbec i realize i need to have a build environment14:17
*** dobee has joined #zope3-dev14:17
philiKONflox, i have some comments about your changes14:17
philiKONi'm writing an email14:17
floxi do not see where i hv problem on the first commit (70042)14:19
philiKONflox, sent14:22
philiKONflox, are you running unit + functionanl tests?14:22
floxi am on Windows, i hv problem running such tests directly on the 'checkout'14:22
philiKONour policy is that all tests need to pass before you check in14:23
philiKONi know you can run tests from a checkout14:23
philiKONon windows14:23
philiKONbecause other z3 developers do it as well14:23
philiKONflox, i think you need to install the .pyd files14:23
floxi have to figure out how i do that14:23
philiKONflox, also, see my email about discussing bigger changes like that first14:24
floxi see14:25
floxi was trying to lower the dependencies between packages14:28
floxjust in case someone want to get rid of deprecated packages14:28
floxnext time, i will consider twice before to commit...14:30
philiKONdon't consider14:31
philiKONa) write proposal for bigger changes14:31
philiKONb) run tests14:31
philiKONwe need to pick our battles14:32
philiKON"just in case someone wants to do X" is not good enough, really14:32
philiKONif you have a use case, that's fine14:32
philiKONbut it seems like you don't14:32
philiKONflox, if you want to help with zope 3, you could fix some bugs :)14:33
floxi do not see many bugs on 3.3 release, these days..14:35
floxi go to lunch. bye14:39
*** stub has quit IRC15:17
*** flox has quit IRC15:26
MJsrichter: ayt?15:28
MJsrichter: you need to check up with philiKON here on those layers :)15:28
*** ktwilight has quit IRC15:28
MJsrichter: See, I don't think that the BB is a bug.. ;)15:28
MJBBB even15:28
philiKONMJ, ?15:28
philiKONah, i see srichter 's post15:29
*** natea has joined #zope3-dev15:29
* philiKON replies15:29
srichterhuh, layers are really deprecated for the page directive?15:30
philiKONwho said that?15:30
srichterok, are we referring to different things?15:30
philiKONthat looks like it :)15:30
srichterAPIDOC says so based on an E-mail I got and just replied to and I thought you guys were referring to.15:31
philiKONsrichter, wait, i know15:31
philiKONLayerField is deprecated15:31
philiKONnot the layer attribute15:31
philiKONonce layerField is removed, it will be an InterfaceField15:31
srichterok, so APIDOC shows this incorrectly, because I bet it uses the layer field still15:31
srichterBBB must be done different for those cases15:32
philiKONwell, it's not apidoc's fault, i guess15:32
philiKONLayerField has a BBB docstring15:32
srichterright, so that approach is a bit too simplistic... I do not have an idea what to do instead though15:32
MJmaybe the BBB warning should be moved out of the docstring?15:32
MJMake it a comment with the object?15:32
philiKONyes, that's what i did initially15:33
philiKONsomebody added the BBB docstring15:33
philiKONof course, that somebody didn't add a good explanation15:33
philiKONwhich is what i usually do in user-relevant texts15:33
philiKONsuch as deprecationwarnings15:33
philiKONor docstrings15:33
philiKONso either move the BBB out o fthe docstring, or extend the docstring15:33
srichterok, I would remove it15:34
srichterbecause we are *not* deprecating the layer field as such15:34
* philiKON is out of check in range15:35
*** ktwilight has joined #zope3-dev15:39
*** flox has joined #zope3-dev15:41
*** ignas_ has quit IRC15:42
*** torkel_ has joined #zope3-dev15:44
*** faassen has quit IRC15:49
*** benji has joined #zope3-dev15:58
*** baijum has quit IRC16:02
*** ignas has joined #zope3-dev16:23
*** gumpa has joined #zope3-dev16:40
*** ofer has quit IRC16:43
*** tarek has joined #zope3-dev16:44
*** batlogg has quit IRC16:44
*** dobee has quit IRC16:52
*** dobee has joined #zope3-dev17:04
*** hdima has quit IRC17:04
*** dobee has quit IRC17:08
*** eins has quit IRC17:08
ignasis there a way to override ++resource++ thingie used in tal templates ?17:15
floxphiliKON: i changed zope.testrecorder, to remove dependency on
*** philiKON has quit IRC17:19
ignasas i want to see the exact code that generates the foo/@@/some-image.png urls17:20
ignasand plug into that machinery17:20
srichterignas: it is a namespace as far as I know, so you can override that17:21
floxignas: see zope.traversing.namespace17:21
srichterignas: but I would strongly argue against plugging into this, because you make a non-compatible change17:21
ignasnon-compatible with what?17:22
srichterwell, then it becomes hard to talk to other people17:23
srichterbecause you have your own thing17:23
ignasi see17:23
ignasthe problem is that i have navigation data stored in url like http://schooltool/+calendar/2001-01-01/daily/persons/peter17:24
ignasat the moment http://schooltool/+calendar/2001-01-01/daily/@@/tab.png works17:24
ignasbut i would rather have resource urls generated without including navigation data in the url17:24
srichterI am really surprised this is done17:25
ignasi have hooked in into IAbsoluteURL for schooltool app17:26
srichterbecause resource URLs should be based on the site, which is schooltool in this case17:26
ignasand resources are using it to get the url17:26
srichterif you say "context/++resource++/tab.png/absolute_url" you will always get "nearestsite/@@/table/png"17:26
srichterwhy are you hooking into IAbsoluteURL?17:27
ignasso that navigation data would be stored in the url without changing all the tal templates, or code of the application17:27
srichterignas: the goal of schooltool should be to become much more of a standard Zope 3 app, not less ;-)17:27
ignasthe goal of schooltool is to become more usable for teachers/students not to become more standard ;)17:28
srichterignas: I would strongly advise against that17:28
srichterignas: then forget building a community around it17:28
ignasthe change is not that harash17:29
srichterignas: while your approach is very REST friendly, I think it is impractical17:29
srichterstoring the state is a session would be the easiest17:29
srichterthe second solution I would suggest is to use a pluggable traverser17:29
srichterthis way you need no hooks into the existing setup17:29
ignasi am using a pluggableTraverser17:30
ignasbut to put the data into URL i have to modify absolute urls17:30
ignasi think modifying urls is way more convenient for the user17:30
ignasas he can use bookmarks17:31
ignaselse the application would become more like gmail17:31
srichterI do not see why you have to hook into that17:31
srichterwe have recently used pluggable traverser to do the same thing in another project17:31
ignassame thing being ?17:31
srichterusing pluggable traversers to describe menu state17:32
ignasyou mean adding fake objects into the __parent__ __name__ stack17:32
srichteranyways, your design decision; though soon noone will be able to help you here anymore, because it is not what everyone else is doing17:32
srichteryou do not have to fake anything17:32
srichtereach traversal step represents a resource, so you can create a proxy object for it17:33
ignashow do your urls look like ?17:33
srichterfor example17:33
srichterall items of the system are in:17:33
srichterit is a regular container17:34
srichterI now want a list of all of my items and get the list overview17:34
srichterSo the main tab "My Items" can be selected and the sub-tab "List" is selected17:35
srichterof course we are using viewlets for the menu17:35
*** alecm has joined #zope3-dev17:38
ignasthis concept adapted to the requirements i have might work17:40
ignaswhile it would be more complex on the programming side17:41
srichterignas: I know your requirements, so I know it works :-)17:41
ignasas navigation data would be easier to lose, while it would be definitely more standard17:41
srichtermaybe you can make some custom ZCML directives and base classes to help you out17:42
ignasthese are code and complexity too17:42
ignasthe state stored this way is very easy to loose ... especially when you have 2/3 of the application is working in another way17:42
srichterbut one written and tested, it is hidden complexity, instead of one that is constantly around17:42
ignasit's not constantly around really17:43
srichterI think that you will find that once you converted the entire view it is much harder to loose the state than you think17:43
srichterignas: I think all of ST will switch to this UI design, otherwise it si pointless and confusing17:44
srichter(I was there when the UI changes were discussed.)17:44
ignasit will, in a few months17:44
ignaswhy do you think having site absolute url modified is such a bad thing ?17:45
srichterit is just a feeling really; it touches too many existing well-working components17:46
srichterthus providing a risk of having unexpected behavior17:46
srichterand having the state stored at the end of the URL feels more natural to me17:46
ignaswith a traverser that eats navigation data and returns the site on the last step it is not touching anything really17:46
srichterwell, as I said, it's more of a feeling, but clearly with you having questions about it and having this resource issue is a sign for me that I am right :-)17:47
ignasthe problem appeared when i tired to make the default context change from the application to something else17:47
ignasi would have different but hard problems if i would try to make site root give out different objects every time too17:48
ignasi just wanted to not have the /persons/manager/ part in the url17:48
ignasnot depending on where is the navigation data after it or before it17:49
ignasas I gather that it will not be actually possible to switch the context with the new style navigation17:49
ignascalendar tab will always point to your calendar17:49
srichterI think that .../calendar/2006-09-08/daily is nice and has all the state you need17:52
srichterViewlets provide you with the flexibility you need here17:52
srichterthe fun thing about viewlet managers is that they can look at parents and whatever to make their decision which viewlet to display and highlight17:52
ignasi didn't really need any viewlets, ok maybe 1 content provider that allows you to display/change navigation data17:53
srichteryou should really become comfortable with them :-)17:59
srichterfor any advanced UI, menus just don't cut it17:59
ignaswhat menus :D17:59
srichterI would have one viewlet manager for the main and the sub tab menu18:00
ignasas for the xomplexity of my changes - the most complex bit was modifying the navigation data in views - creating a url that would keep the mode and tab components while changing the date etc.18:00
*** faassen has joined #zope3-dev18:00
srichterI use that pattern in another app and it works wonderfully, because the menu is totally independent of the context18:00
ignaspretty much all of the structural changes went into
ignasby the way did you know that zc datewidget doesn't really work with virtual hosting18:04
ignasat least in schooltool18:04
ignasas it accesses resources using "/@@/zc.datetimewidget/datetimewidget.js" instead of the full URL18:04
faassenignas: that might be a zc.resourcelibrary issue then?18:05
ignasmaybe, vidas was working on it and he said that he could not find the place where these urls are generated18:05
faassenignas: I'm not sure whether zc.datetimewidget is using the resourcelibrary but it's a good chance. were they <link> urls in the HTML head>18:06
faassenignas: as zc.resourcelibrary does put them in18:06
ignasand they assume that the application will not be hosted in a server under some directory like
ignasyes they were <link> urls18:06
faassenhm, that sucks.18:06
faassenis where the links get put in.18:07
faassenyou'd think it would work as it should be getting the site root's URL.18:08
faassen            site = getSite()             baseURL = str(getMultiAdapter((site, self._request),                                           IAbsoluteURL))18:08
faassenanyway, hopefully that helps you. zc.resourcelibrary is actually quite useful.18:09
faassenbecause you don't have to worry about including links in your own pages anymore.18:09
faassenthey just get included when you need them, say, somewhere deep inside a widget.18:09
ignaswell zc.library egg 0.5.1 seems to not have the part that adds base_url18:11
faassenso this was a later bugfix18:11
ignasand that file hasn't got it's svn tags set up properly :P18:11
faassenyes, three weeks ago.18:12
faassenwhich file?18:12
faassenwhat do you mean with svn tags set up properly?18:12
ignasas svn revision headers $Id$18:12
ignasare the same on both of them18:12
faassenyou mean it doesn't have revision header replacements?18:12
ignasyes, wrong word18:12
faassenanyway, so the fix would be to pull a new egg of zc.resourcelibrary, most likely. :)18:13
ignasnot tags18:13
ignasfaassen: where is the version set and which one is the most up to date ?18:13
*** zagy has quit IRC18:13
faassenignas: what do you mean?18:14
ignasnp, just saw that it is tagged as 0.5.218:14
faassenthe latest release appears to be 0.5.218:14
faassenignas: oh, I see some schooltool packages made it into ZPL land?18:15
faassenignas: like the pluggable traversers bit.18:16
ignasoh, cool :)18:16
faassenignas: didn't know we'd gotten permission to relicense that stuff.18:16
faassenignas: there's some useful stuff, though a bit underdeveloped, for local utility registration, in schooltool.utility18:16
faassenthere's so much cool stuff in that now. :)18:17
ignassrichter: i will seriously look into the moving of navigation data to the end of the url to avoid touching of the absolute url adapter18:18
ignasthough probably i'll have to postopone the migration at the moment18:19
ignasmaybe i should just leave a string component in the url just to keep the part "site url should point to the site". Would it make sense to have a traversal step from application that would point to the currently logged in user ?18:24
srichterignas: Fielding would say that pointing to the current user is unrestive :-)18:30
srichteryour approach with putting the state into the URL is more appropriate18:31
*** jukart has quit IRC18:36
*** J1m has joined #zope3-dev18:47
faassenJ1m: hey18:48
faassenJ1m: I'm thinking about buildout a bit more, do you have time to hear some thoughts?18:49
*** markup_ has joined #zope3-dev18:55
faassenoh, I see you already thought about it (of course)18:56
faassenJ1m: I'm having an application that wants to install instances. I have the opportunity to write the instance creation logic as part of the recipe.18:56
faassenJ1m: but that means the instance creation logic is actually in the recipe and the application becomes a library.18:57
faassenJ1m: and I wonder what that means. Like, a config file template would go into the recipe.18:58
*** zagy has joined #zope3-dev18:58
faassenthough I imagine eggs have a way to store data.18:58
faassenand to maintain the code? hm.19:01
*** xenru has joined #zope3-dev19:01
*** MJ has quit IRC19:03
*** schwendinger has quit IRC19:16
*** projekt01_ is now known as projekt0119:20
*** alecm has quit IRC19:23
*** schwendinger has joined #zope3-dev19:27
J1mfaassen: I'm here now19:27
faassenJ1m: okay, I'm still pondering issues.19:28
J1mso, obviously, the zope3instance recipe creates instances.19:28
faassenJ1m: basically I have a situation much like the zope 3 library (which theoretically could become an egg/eggs, even though now you have a z3checkout recipe)19:28
J1mIt's logical, in my mond, to treat instances as parts.19:28
faassenJ1m: and an instance.19:28
faassenin your mond?19:29
faassenoh, mind.19:29
faassenyeah, sure.19:29
faassenanyway, let me sketch out the situation first.19:29
faassenI mean, it's analogous, but simpler and more difficult at the same time as I don't have something like a mkzopeinstance script yet.19:29
faassenso, I install the package as an egg.19:30
faassenbut then I want to make an instance for this package.19:30
faassenthis has a bin dir, a log dir, an etc dir with config files, and a var dir.19:30
faassenso very much like a zope 3 instance.19:30
faassenso what I could do is write a recipe that make all that, dumping the default config file in there, etc.19:30
faassenI started doing that, but I started thinking about where I want to keep the 'skeleton'19:31
faassenand then there's also the facility eggs provide for script entry points.19:31
faassenand package data. but this isn't really package data.19:31
faassendoes one maintain the skeleton for an instance inside the recipe?19:31
d2msrichter: http://localhost:80808/++apidoc++/Code/zope/formlib/ is missing tests, ftest and page from the listing -- do you need a mail to mailing list ?19:31
J1mremond me to mention our experience with the various instance directories. :)19:32
faassenwhat about the use case of running the instance from the checkout directly? I'm doing that now.19:32
faassenJ1m: you like typing mond instead of mind today. :)19:32
faassenJ1m: anyway, so I was wondering what way to go.19:32
faassenJ1m: when generating the instance, I'd like to plug in some stuff, like fill in some config data that I can deduce from another part.19:33
* J1m hates skeletins 19:33
* faassen grins.19:33
*** mgedmin has joined #zope3-dev19:33
J1mYeah, so a recipe may have various templates that help with generation of instance files.19:33
faassenand they hate us, hiding in closets and then suddenly coming out, and there are whole armies too, of these skeletons.19:33
J1mThese can be in the recipe package.19:33
faassenanyway, if you hate skeletons, what would you recommend instead?19:34
J1mThey can be much more modest than the classic zope skels.19:34
faassenyes, I have a pretty modest skeleton.19:34
J1mSo first of all, Zope instances classically have way too much structure, IMO.19:34
faassenbasically a few empty directories, and a few tiny scripts and a config file.19:34
faassenanyway, I do have the usecase thatI want to mess with both LD_LIBRARY_PATH19:35
J1mAll these silly directories are just annoying.19:35
faassenas well as Python path.19:35
J1mI'll come back to LD_...19:35
faassenyou'd recommend keeping it flat and just putting config and data in the same directory?19:35
faassenand log file?19:35
faassenand scripts.19:35
*** romanofski has quit IRC19:36
J1mSecond, when we actually deplot things, our customers (meaning our SAs) want things installed in the classic Unix way.19:36
*** romanofski has joined #zope3-dev19:36
J1mThey want log files in /var/run19:36
J1mI mean /var/log19:36
J1mThey want pid files in /var/run19:36
faassenyeah, and scripts in something/bin19:36
J1mand so on.19:36
J1mEventually, they may want me to scatter the scripts and config files too.19:37
*** schwendinger has quit IRC19:37
J1mSo, in our old buildout sw, we have strategies for spreading this stuff around.19:37
J1mI expect to start introdcing some suppoer for this in the current buildout.19:37
J1mnew buildout.19:37
J1mI expect, for example, that the buildout will grow log-directory, pid-directory, maybe even config-directory options.19:38
faassenanyway, one thing that bothers me a bit is that the recipe has like, *the* config file template.19:38
J1mwhere these will default to buildout subdirectories and in deployment will point to system directories.19:39
faassenright, that'd be useful.19:39
J1mI'm not sure what your point is in referring to "the" config file template.19:39
faassenwell. it gets spread around.19:40
faassenthe recipe knows the config file, has the basic example.19:40
J1mhow so?19:40
faassenand the software, the egg, accesses it.19:40
faassenwhat if you want to do integration tests?19:40
J1mNo, the recipe accesses it to create the config file that the sw uses.19:40
faassenthat's true.19:41
faassenso what I could do is maintain a skeleton in the recipe's package19:41
faassenso it actually comes with the egg.19:41
faassenand then use the egg data access API to talk to it.19:41
faassenand copy it into the instance dir.19:41
faassenand modify it to push in the right config things (based on another part we already have)19:42
faassenand also puts in the scripts, and modifies those, messing about with the Python path.19:42
faassenand possibly LD_LIBRARY_PATH at all, if that's possible from within a python script.19:42
*** projekt01 has left #zope3-dev19:42
faassenat all -> as well.19:42
*** volvox has quit IRC19:42
J1mI can usually use a distutils option to avoid LD_LIBRARY_PATH.19:43
faassenLD_LIBRARY_PATH actually needs to have added to it a directory in another art.19:43
faassenwould that be doable with a distutils option?19:44
J1mBTW, did you notice the zc.recipe.egg:custom recipe?19:44
J1mHere's a sample part section that builds the spread module:19:44
J1mrecipe = zc.recipe.egg:custom19:44
J1megg = SpreadModule ==1.419:44
J1mfind-links =
J1minclude-dirs = ${buildout:directory}/parts/spreadtoolkit/include19:44
J1mlibrary-dirs = ${buildout:directory}/parts/spreadtoolkit/lib19:44
J1mrpath = ${buildout:directory}/parts/spreadtoolkit/lib19:45
faassenah, okay.19:45
J1mSo, the custom recipe lets you build an egg with custom distutils options.19:45
faassenah, useful.19:45
J1mThe rpath option lets you control where dynamic libraries get looked for,19:45
faassenokay, so I'd need to set rpath.19:46
faassenspecify a custom egg for..19:46
faassenhm in my case..19:46
faassenhm, I wonder whether that works in my case.19:47
J1mIn my case, I wantd the buildout to control the spread install.19:47
faassenI'm not actually building anything.19:47
faassenI have the following setup..19:47
J1mAnd therefore, I had to tell the spreadmodule not to look on /usr/local.19:47
faassenan openoffice recipe, which downloads an openoffice and installs it, messing with its Python so it uses the buildout python (in the future configurable python)19:48
faassenand there's an oooconv egg, which contains the sources to software to drive openoffice as a server.19:48
faassenand then a oooconv recipe, that installs an instance of oooconv19:49
faassenand when the instance's script runs, the LD_LIBRARY_PATH needs to be set to some direcory in the openoffice part.19:49
J1mah, so you might be stuck with hacking LD_...19:49
J1mtoo bad.19:49
J1mYou'll have to generate a shell script to set that. :(19:50
faassensetting up openoffice in server mode with Python is an enormous pain.19:50
faassenthat's why I'm making a buildout.19:50
faassenyeah, I think I'll do that.19:50
faassenanyway, the way is reasonably clear.19:51
faassenthanks for thinking this through with me.19:51
faassenI was grasping for the right way to go about this. anyway, I'll try this out.19:51
*** romanofski has quit IRC19:56
*** romanofski has joined #zope3-dev19:56
*** ignas has quit IRC19:56
*** kobold has left #zope3-dev20:07
d2msrichter: http://localhost:8080/++apidoc++/Code/zope/app/traversing is missing too, looks like you are testing for deprecated modules ? is this intentional ?20:08
nourifaassen: Your discussion with Jim reminds me of the Shell script we had working on Wednesday or so :)20:11
nourifaassen: Of course it doesn't solve all problems, just noting20:11
mgedminif I use defineChecker in a unit test's setUp, is there a way to undefine it in the tearDown?20:14
*** alecm has joined #zope3-dev20:15
*** natea is now known as natea|away20:15
*** RaFromBRC has joined #zope3-dev20:17
*** jinty has quit IRC20:20
faassennouri: it doesn't set the LD_LIBRARY_PATH, you need to run your env script.20:25
*** torkel_ has quit IRC20:25
nourifaassen: it includes the env script, but yeah, if you want to start up the server you need env20:26
* nouri -> dinner20:26
*** jukart has joined #zope3-dev20:37
*** zagy has quit IRC20:54
*** batlogg has joined #zope3-dev20:56
*** MJ has joined #zope3-dev20:58
*** schwendinger has joined #zope3-dev21:00
*** natea|away is now known as natea21:04
*** zagy has joined #zope3-dev21:10
*** tarek has quit IRC21:14
*** alga has joined #zope3-dev21:19
*** faassen has quit IRC21:42
*** mkerrin has quit IRC21:58
*** projekt01 has joined #zope3-dev22:02
J1mmgedmin: yes22:11
J1mThere is an undefineChecker method.22:11
mgedminah, thanks22:11
mgedminoh, and CleanUp also cleans them22:14
* mgedmin happy22:14
*** jukart has quit IRC22:25
*** timte has quit IRC22:29
*** jinty has joined #zope3-dev22:52
*** schwendinger has quit IRC23:08
*** jinty has quit IRC23:22
*** dunny has joined #zope3-dev23:26
*** MJ has quit IRC23:36
*** MJ has joined #zope3-dev23:37
*** mgedmin has quit IRC23:44
*** ignas has joined #zope3-dev23:44

Generated by 2.15.1 by Marius Gedminas - find it at!