IRC log of #zope3-dev for Wednesday, 2005-02-16

mgedminok, I did that00:00
mgedmincan someone test whether -1 works again?00:01
mgedmin(my ftests unconditionally turn on that feature, so I see no change with -1 or without it)00:01
srichterI will check it tonight00:01
srichterbtw, stub forgot to check in pytz, so my changes have to wait till tomorrow00:02
*** hazmat has quit IRC00:11
*** _projekt01 has quit IRC00:11
*** alga has quit IRC00:16
*** Workblia is now known as Awayblia00:21
*** jhauser has joined #zope3-dev00:23
*** GaryPoster has quit IRC00:31
*** mgedmin has quit IRC00:33
*** jhauser_ has quit IRC00:35
*** jhauser has quit IRC00:49
J1msrichter, ayt?01:06
srichterJ1m: yep01:08
J1mCode that constructs ServiceRegistrations is screwed01:09
J1mThey no-longer support the old constructor signature01:09
srichtermmh, I was hoping that noone would do this by hand01:10
J1mwell, it used to be necessary to add things like local utility services when you created sites01:10
srichterfor tests there is setup01:10
J1mWe have code that creates and initializes sites.01:10
J1mI suspect that others do too01:10
srichterand you do not use setup, ztapi, etc. to do this?01:11
J1mI'm willing to assume that we are weird.01:11
J1mThis is not test code01:11
srichteroh really?01:11
J1mWe allow someone to add a special site01:11
srichterI see01:11
J1mThat special site gets configured as it is added.01:11
srichterreally, a lot of the test API is useful for this kind of stuff01:12
srichterit should really move to another location01:12
J1mIf I remove the code to add the utility service, I get farther01:12
J1mThen a call like:01:12
srichterok, I look into this tonight01:12
J1mutilities = sm.queryLocalService(zope.component.servicenames.Utilities)01:12
J1mfails with:01:13
J1mComponentLookupError: (<InterfaceClass zope.component.bbb.service.IService>, 'Utilities')01:13
J1m(<InterfaceClass zope.component.bbb.service.IService>, 'Utilities')01:13
srichteroh, darn the error there should be caught of course01:14
J1mIt should?01:14
srichterfeel free to shoot me personal emails01:14
srichterit is a query* call01:14
srichternote that queryLocalService looks only for local services and will not find the global one'01:14
J1mBut should it fail?01:14
J1mdoesn't a local site have it's own utilities service?01:15
J1m(or the equivalent)01:15
J1mdoesn't a local site have it's own utilities registry?01:15
srichteroh, of course01:15
srichterin the new API it does of course01:15
srichterI probably forgot to add a special if statement for the utility and adapter service01:15
srichterI know I did it for zope.component01:16
srichteryes, it does01:16
J1mSo, are the registries equivalent to the services?01:16
srichterevery site manager has adapters and utilities01:16
J1mBut do these things it has behave like the old services?01:16
*** benji_york has quit IRC01:16
srichterthey pretty much have the same API, yes01:16
J1mI really want to get our stuff working again, but I'm afraid that if I just fix our stuff, you'll run into this with someone else later.01:18
srichteryeah, we should have never supported getting the utility service and then lookup a utility this way01:18
srichtergetUtility() from zapi is so much nicer01:18
srichterfortunately, I fixed that badness in the trunk01:18
J1mIn this case, there was code checking to make sure the utility service was present.01:19
srichteryeah, this is a valid reason01:19
J1mwell, it was01:19
J1mIt shouldn't be necessary any more01:19
srichternope :-)01:19
srichterwell, I think once you and Roger reported their problems, we should be pretty ok01:20
J1mI'm not exactly sure how to report them.01:20
J1mI'll send you some email01:20
srichterI am a bit afraid of Roger's reports, since he reused a lot of browser code; I have not even attempted to provide BBB for this01:21
J1mwhat do you mean browser code?01:21
srichterJust say, this API must have changed... like you did here on IRC01:21
srichterI cannot test it anyways01:21
srichterpresentation classes == browser code01:21
J1mI still don't know what you mean01:22
srichterlike the tools classes01:22
srichterLet's say that someone hypothetically implemented their own custom tool type01:22
srichterthey would have used some base classes from
srichterfor things like that we have no BBB01:23
srichterbut I think that relying on classes in BB is not a good idea anyways, except for some exceptions like
J1mOf course, but the whole point of backward compatibility is that we need to support old code.01:24
J1mOf course, we should only support old "supported" code.01:25
J1mAnd we haven't solved the problem of making it clear what is supported/public. :(01:25
srichterit gets back to my point that we need a way to state what interfaces/classes/functions are public and supported01:27
srichtermaybe some flag would be find01:27
srichter__public__ = True01:27
J1mI have suggested in the past that an api is public if it's documented in an or an interfaces package.01:29
J1mI think that that is still workable.01:29
J1mAnything else is non public.01:29
J1mWe'll need to discuss this on the list I think.01:29
srichterright, and make it a guideline01:30
J1mIt would need to be more than a guidline01:30
J1mIt would need to be a contract01:30
*** tvon has joined #zope3-dev01:34
J1mI'm pretty sure the LocalUtilityRegistry doesn't implement ILocalUtilityService01:45
J1mI'm guessing that a LocalSiteManager implements much of the old ILocalUtilityService.01:47
*** mexiKON has joined #zope3-dev01:50
J1mExcept that it doesn't implement enough to be backward compatible.01:52
J1mfor example, it doesn't implement getLocalUtilitiesFor01:53
J1mBut that's not a public api01:59
*** philiKON has quit IRC01:59
J1mso, not a problem there01:59
srichterok, time for me to go home02:01
srichtersee ya later02:01
*** srichter has quit IRC02:01
*** mgedmin has joined #zope3-dev02:01
*** Aiste has quit IRC02:02
*** hazmat has joined #zope3-dev02:11
*** d2m has quit IRC03:07
mgedminI ran ./ -pv zope in my Zope3 checkout, and ~7 minutes later my system started swapping03:12
mgedminthe python2.3 process that ran ate 409 megs of virtual memory03:13
mgedminany ideas why?03:13
J1mdoes it happen if you run it again?03:13
mgedminI'll try03:14
J1mSounds vaguely familiar.  Like maybe related to a bug Tim fixed in ZODB, but I'm not sure.03:14
J1mI'll run the tests and watch w top03:14
mgedmin100 megs already03:15
mgedmin(still at 24% of unit tests)03:15
mgedminoops, wrong python2.303:16 is at 86 megs (45% of unit tests)03:16
*** hazmat has quit IRC03:16
mgedmin147 megs after 100% of unit tests03:18
mgedmin190 megs at 33% of ftests03:19
J1mAre you on a 64-bit machine?03:19
mgedmin245 megs at 47% of ftests03:20
mgedminno, it's a pentium-IIIM laptop03:20
mgedmin310 megs at 56% ftests03:20
mgedminI'm running ubuntu hoary03:20
J1mw python2.3.4, I got up to about 125 megs w the unit tests03:21
J1mI guess I'm seeing about the same memory usage03:21
J1mWe realy need to spend some time looking for leaks03:22
mgedmin390 megs at 93% ftests03:22
J1mYup, same for me03:22
J1mmore or less03:22
mgedminram usage stayed pretty constant for the first 20% of ftests03:22
mgedminbut started growing rapidly after that03:23
* mgedmin shrugs03:23
J1mPerhaps we need a disk-based mapping storage :)03:23
* J1m shrugs too03:24
J1mWe realy need to spend some time looking for leaks :)03:24
J1mwe have some pretty good tools for doing so03:24
J1mI have all of my tests passing now. Yay!03:24
*** J1m has quit IRC03:25
*** hazmat has joined #zope3-dev03:36
*** hazmat has quit IRC03:54
*** srichter has joined #zope3-dev04:07
*** ChanServ sets mode: +o srichter04:07
*** hazmat has joined #zope3-dev04:42
*** mgedmin has quit IRC05:00
*** `anthony has quit IRC05:11
*** hazmat has quit IRC05:23
*** hazmat has joined #zope3-dev06:00
*** hazmat has quit IRC06:18
*** `anthony has joined #zope3-dev06:41
*** xenwryly has joined #zope3-dev07:14
*** jhauser has joined #zope3-dev07:14
xenwrylyhello all07:15
xenwrylyis there anyone out there working on any other Z3 skins other than Rotterdam, ZopeTop and Basic?07:15
*** xenwryly has quit IRC07:33
*** sashav has quit IRC07:58
*** SteveA has joined #zope3-dev08:49
*** apauley has joined #zope3-dev09:10
*** sashav has joined #zope3-dev09:16
*** tvon has quit IRC09:27
*** `anthony has quit IRC09:28
*** d2m has joined #zope3-dev09:41
*** hazmat has joined #zope3-dev09:45
*** stub has joined #zope3-dev09:47
*** zagy has joined #zope3-dev09:52
*** bradb has joined #zope3-dev10:01
*** tvon has joined #zope3-dev10:04
*** Theuni has joined #zope3-dev10:11
*** AJC has joined #zope3-dev10:48
*** SteveA has quit IRC10:51
*** sashav is now known as sashav|afk10:52
*** SteveA has joined #zope3-dev10:54
*** MalcolmC has joined #zope3-dev11:06
*** sashav|afk is now known as sashav11:41
*** tarek_ has joined #zope3-dev11:53
*** SteveA has quit IRC12:11
*** SteveA has joined #zope3-dev12:15
*** mooded has joined #zope3-dev12:49
*** SteveA_ has joined #zope3-dev13:06
*** SteveA has quit IRC13:12
*** SteveA_ is now known as SteveA13:12
*** hazmat has quit IRC13:16
*** hazmat has joined #zope3-dev13:44
*** J1m has joined #zope3-dev13:54
*** `anthony has joined #zope3-dev14:01
*** projekt01 has joined #zope3-dev14:09
*** J1m has quit IRC14:25
*** srichter has quit IRC14:28
*** niemeyer has joined #zope3-dev14:28
*** SteveA has quit IRC14:31
*** SteveA__ has joined #zope3-dev14:33
*** SteveA__ is now known as SteveA14:33
*** regebro has joined #zope3-dev14:43
*** hazmat has quit IRC15:05
*** srichter has joined #zope3-dev15:26
*** ChanServ sets mode: +o srichter15:26
*** mgedmin has joined #zope3-dev15:28
*** sashav has quit IRC15:57
*** SteveA has quit IRC16:00
*** SteveA has joined #zope3-dev16:15
*** regebro has quit IRC16:17
srichtermgedmin: I just checked in a lot of goodies for you. :-)16:27
srichtermgedmin: zope.i18n.format should now be ready and useful for your calendaring app.16:28
srichterlet me know, if there are other issues left16:28
mgedminsrichter, thanks!16:28
mgedminI suppose I should write about schoolbell in the mailing list16:29
mgedminwe're not just building a calendaring app; we also want to build a reusable zope 3 calendaring library16:29
srichteryeah, you should16:30
srichterI think we really need a newsletter again16:30
srichterI am so happy that we have timezone support now16:31
srichternow, I just have to get rid of this stupid datetimeutil package and reimplement its functionality using zope.i18n.format16:31
*** mooded has quit IRC16:40
SteveAmgedmin: happy independence day16:44
*** apauley has quit IRC16:46
* Theuni just saw "hula".17:03
*** SteveA_ has joined #zope3-dev17:04
srichterTheuni: huh?17:04
Theunithe novel netmail server which was just released open source17:10
Theuniamazing thing17:10
Theuniintegrated mail/calendar solution. comes with own smtp/ldap/imap/pop/web server, is pretty fast and flexible17:10
*** SteveA has quit IRC17:14
*** tarek_ is now known as tarek_away17:24
*** AJC has joined #zope3-dev17:31
*** J1m has joined #zope3-dev17:32
srichterJ1m: do you have a moment?17:32
srichterJ1m: so due to the removal of services and special presentation components, there is some opportunity and need to change the API doc tool before 3.117:33
srichterany ideas about possible improvements?17:33
srichterI already remvoed the services section, of course17:34
* mgedmin perks his ears17:34
srichtermgedmin: sure any commets are welcomed17:34
srichterfor now I left views and adapters separate17:35
J1m1. Show all adapters. Don't filter out views.17:35
srichter(I detect views by checking whether the last required interface isOrExtends IRequest)17:35
J1mGiven that the view apis have been deprecated, we might as well show everything together,17:35
srichternote that for views I can show many interesting things that are not applicable to regular adapters17:36
srichtersuch as the page template17:36
J1m2. When showing adapters (for an interface), break them out by the interfaces they were registered for,17:36
srichterand the original class, because for views the class the adapter provides is not the one you specified17:36
J1mFor example, separate the adapters registered for * (default) from adapters registered for gneral interfaces, from adapters registered for the specific interface.17:37
J1m3. Show or ptovid ea way of easily seeing the ZCML used.17:37
J1mShowing zcml should include the file path and position.17:37
J1mNote that if you showed the zcml, then people can see things like the template used.  You don't nee special view-handling code.17:38
J1m3. Show or provide a way of easily seeing the ZCML used.17:38
srichtertrue, but I do not know whether I like it ;-)17:39
J1mlike what?17:39
srichter(but I am writing everything down first)17:39
srichtera view section also allows me to sort views by skin17:39
srichteror layer actually17:39
J1mIf you want to sperate the views, then at least show the views for an interface.17:40
J1mIf you want to seperate the views, then at least show the views for an interface.17:40
J1mWhen looking at an interface, I should be able to see all of the adapters, including views immediately.17:40
srichterok, I am not doing this yet?17:40
srichterOkay, I think I can get rid of most of the view documentation module then and move its funcitonality to the interface code17:41
srichterI like that17:42
J1mNo, you are not doing this yet.17:42
srichterso basically you can look for specifications using classes and interfaces17:42
srichterok, then this was an oversight from the beginning17:43
srichterany ideas about navigation?17:43
srichterI really do not like the tree view of the interfaces anymore17:43
J1mIt would be nice to gave a type type module for listing the meta types.17:43
srichterI think I did a better job with the class search interface and browser17:43
srichterJ1m: sorry, I did not understand the last sentence (too many types ;-)17:44
J1mIt would be nice to have a type type module for listing the meta types.17:44
srichterwhat types? what meta types?17:44
J1mYou get a list of the type types (e.g. ContentType, ToolType, etc.17:44
srichterohh, I see17:45
J1mYou expand a type type and get all of the typs of that type.17:45
srichterthis is a cool idea17:45
J1mMaybe call these interface types.17:45
J1mI like the interface tree17:45
srichterI think this will be very helpful17:46
J1mI don't like the search interface.17:46
J1mIt would be useful if the search was a text search17:46
J1mBut a python path search isn't useful to me.17:46
*** tvon has quit IRC17:47
J1malthough, maybe I didn't know how to use it correctly.17:47
J1mFor search, it would be nice to search based on text.17:47
J1mSomeday down the road, we should ahve a tool that uses a full-text index.17:48
srichterI think the Python path is still useful17:48
J1mSomeday, it would be nice to have a tool that reads a site configuration and computes a separate database that could be browsed by a separate application.17:49
srichterbut I will add an option for that17:49
J1mIt's a bit annoying to have to have zope running to use the tool.17:49
J1mAnd, of course, we don't want apidoc enabled in a production server.17:49
srichterI hope that we eventually will provide an online version17:50
J1mI really like apidoc a lot.17:50
*** upsidedown has joined #zope3-dev17:50
J1mOK, is that enough input for now? :)17:50
srichterI actually have not used it much at all (how ironic)17:50
srichteryes, this gets me well-started17:51
srichtermgedmin: any comments?17:51
J1mI'm using it more and more.17:51
J1mIt's very handy to review how to use zcml.17:51
srichterI think in 3.1 it will be much more usefull, because 3rd party packages can be registered for the class browser17:51
J1mIt's very handy to see what utilities their are (e.g. permissions)17:51
srichteryes, I think that showing the security information was a good idea; I like this info a lot too17:52
mgedminsrichter, I'd be really happy to have a browseable HTML zcml reference to debug zope 3 startup failures17:52
J1mIt's very nice for borwsing interfaces or classes that live in deeply nested inheritence hierarchoes, because it provides a flattened view.17:52
mgedmincan't really use apidoc if zope doesn't start because of a zcml error ;)17:53
srichtermgedmin: right, eventually, I hope we will have :-)17:53
*** J1m is now known as J1m|away17:53
mgedminI haven't used apidoc for other purposes so far -- usually a recursive grep through all zcml files on an interface or view name is easier for me17:54
srichterJ1m|away: maybe we could set this up, once 3.1 is released?17:54
srichtersame here; Tom was very interested in apidoc as a way to document SchoolTool17:54
srichterI really should think about a way to generate PDF files17:55
mgedminsrichter, use reportlab17:55
srichterI could do that or I just create TeX and compile it17:55
srichterthis way I can use TeX files to define the look and feel17:55
mgedminthat would work too, I suppose17:56
* srichter is a big fan of TeX17:56
mgedminalthough for most users, I suspect, it would be much easier to install reportlab than to install TeX17:56
srichtertrue, but they can use TeX2Word or somesuch tool17:56
srichterof course, we could also generate docbook, so we will have less control of the layout then17:57
mgedminsuppose I have a generic library like schoolbell.relationships17:57
mgedminwhat would be a good way to produce nice developer documentation for it?17:58
mgedminpydoc? epydoc? apidoc?17:58
srichterhold on17:59
srichterso apidoc does not handle README.txt files yet; that's another thing I want to finish for 3.118:00
J1m|awaysrichter, we really need to get 3.1 out18:00
* mgedmin nods18:01
srichterI am working on it18:01
mgedminwhat is holding up 3.1, by the way?  is doc/TODO.txt up to date?18:01
srichterthat's the reason I started going through the bugs18:01
srichtermgedmin: no, some tasks can be removed and later bugs need to be added18:01
srichterbut it is adjusted to 3.1 already18:02
srichterthere are many new XXX comments in src/zope that need to be addressed18:02
srichterI just looked through the newly reported bugs and most of them are not shallow!18:03
srichterso I think the biggest bottleneck right now are bugs and XXX comments18:03
srichterand I do think that having a useful API doctool is as important as fixing bugs and XXX comments18:04
mgedminI think I know how to fix
mgedmin(binding to specific network interfaces)18:05
srichteroh cool18:05
mgedminthe hard part is writing unit tests for ServerType and ServerTypeFactory18:05
mgedmin(there are none at the moment)18:06
srichtermmh, maybe writing some non-integrative unittests would suffice18:06
srichtermgedmin: back to your documentation question; currently apidoc tool is your best bet, since it integrates interfaces and ZCML into the documentation18:07
mgedminI agree18:09
*** juka has joined #zope3-dev18:11
*** upsidedown has quit IRC18:11
*** juka has left #zope3-dev18:13
*** vlado has joined #zope3-dev18:15
mgedmina question regarding ut.py18:19
mgedminshouldn't it be changed so that it runs unittest.main(defaultTest='test_suite') at the end?18:20
mgedmininstead of plain unittest.main()18:20
mgedminI'll do that then18:20
* mgedmin discovered the -s option of and is loving it immensely18:31
srichteryes, I always use --dir18:32
srichterit makes test runs much faster18:32
*** SteveA__ has joined #zope3-dev18:32
*** SteveA_ has quit IRC18:38
mgedminIServerType.create does not match the signature of ServerType.create18:39
mgedminI will assume the interface is out of date18:39
srichterI think this is a good guess :-)18:39
*** vlado has quit IRC18:42
projekt01Stefan, I fixed the blow service branch migration in Tiks.18:44
*** tvon has joined #zope3-dev18:48
srichterprojekt01: oh cool; any hurdles?18:52
projekt01stefan: no, just replace the deprecated methods and change some classes.18:55
*** deo has joined #zope3-dev19:02
*** mooded has joined #zope3-dev19:12
*** tvon|x31 has joined #zope3-dev19:19
*** tvon has quit IRC19:19
*** tvon|x31 is now known as tvon19:20
mgedminzconfig ip-address type question19:21
mgedminif I have a zconfig key of type ip-address, and I say19:22
mgedmin   listen
mgedminI think I will get a tuple ('', 8080)19:22
mgedminas the value for that setting19:22
mgedminwhat will I get if I specify only 'listen 8080'?19:22
mgedmin(None, 8080) or ('', 8080)?19:22
mgedminnever mind, I got the answer19:23
mgedminif sys.platform[:3] == 'win', I will get ('localhost', 8080)19:24
mgedminotherwise I will get ('', 8080)19:24
mgedmin<sarcasm>how nice</sarcasm>19:24
moodedhey all.. i know this is zope3-dev, but here are active & skilled ppl ;) maybe anybody knows / remembers how one must import/instantiate  classes that are not avail in zope normally (lying somewhere else in filesystem) from zope so that they can reference back into zodb  context ?  i have a class derived from ObjectManager, Item that serves a DTMLFile... all can be rendered fine except REQUEST and any call referencing to zodb-objects.. any hints ? if not o19:26
moodedr really wrong channel, sorry for disturbance..19:26
*** tvon has quit IRC19:28
*** tvon has joined #zope3-dev19:35
*** MalcolmC has quit IRC19:43
* mgedmin hits a snag19:47
mgedminit's a bug in ServerBase!19:48
*** jhauser has quit IRC19:51
*** SteveA__ is now known as STeveA19:51
*** STeveA is now known as SteveA19:51
* mgedmin fixes the bug19:57
mgedminI just finished fixing
srichterthanks; I close it20:07
*** zagy_ has joined #zope3-dev20:14
*** zagy has quit IRC20:14
*** tarek_away is now known as tarek_20:16
tarek_Do you, people, have some coverage scripts solution on your tests, using zopectl ?20:16
tarek_I am trying to set up something to list code that is not tested20:17
mgedmin./ has an option to produce coverage reports20:20
mgedminI have no clue what zopectl has to do with your question20:20
mgedmincan zopectl be used to run unit tests?20:20
tarek_that's what i use20:21
mgedminI never knew that ;)20:21
tarek_i run : zopectl test -vv --dir /your/tests/20:21
mgedminI assume it accepts the same set of options as test.py20:21
tarek_you must be right20:21
tarek_it's a -c flag or something right ? i am going to take a look. thanks to you20:22
mgedmintry zopectl test --help and look for 'coverage'20:22
mgedminthe option is -T or --trace20:22
mgedminlook for output in a subdirectory named 'coverage'20:23
tarek_no coverage told on --help20:23
*** sashav has joined #zope3-dev20:24
*** hazmat has joined #zope3-dev20:25
*** tarek_ is now known as tarek_away20:37
*** tvon has quit IRC20:43
*** tvon has joined #zope3-dev20:46
*** zagy_ has quit IRC20:59
*** Theuni has quit IRC21:07
Damascenetal:define="actions context/@@view_get_menu/wikipage_actions" <== i copied this from the wiki product.  i'm trying to use something similar in my product but instead i get a "page not available" error with no apparent errors in the logs.21:13
Damascenei created a configure.zcml file with menuitem entries that refer to that name as well.21:14
Damascenei'm assuming my configure.zcml is correct as it didn't appear to die on startup.  so that seems like the correct tal, but insertion of it gives very vague errors.21:15
*** tvon has quit IRC21:21
*** zagy has joined #zope3-dev21:25
*** tvon has joined #zope3-dev21:27
Damascenehm.  weird.  seems to be working now heh.21:33
Damascenealthough i wonder if this is the solution i really want.  if it's tied too deep with zcml that might be annoying to work with hm.21:43
*** tvon has quit IRC21:53
*** tvon has joined #zope3-dev21:53
*** niemeyer has quit IRC22:07
*** stub has left #zope3-dev22:08
*** hazmat has quit IRC22:10
*** stub has joined #zope3-dev22:11
*** hazmat has joined #zope3-dev22:12
*** bradb has quit IRC22:13
*** tvon has quit IRC22:17
*** stub has quit IRC22:26
*** tarek_away is now known as tarek_22:28
*** bskahan has joined #zope3-dev22:47
*** tvon has joined #zope3-dev22:53
*** d2m has quit IRC23:01
*** d2m_ has joined #zope3-dev23:01
*** d2m_ is now known as d2m23:01
*** sashav has quit IRC23:14
srichterprojekt01: btw, your ideas about the OnlineHelp sound good23:57
srichterprojekt01: but what did you mean by tree status info of the tree and making the tree independent of topic views?23:58

Generated by 2.15.1 by Marius Gedminas - find it at!