*** fdrake has joined #zope3-dev | 00:00 | |
fdrake | there's a test failure on the Zope 3 trunk, w/ a clean checkout | 00:00 |
---|---|---|
*** sm-afk has quit IRC | 00:01 | |
srichter | benji_york: the final speed is even worse: the test before conversion ran steadily at 8 secs, now it needs 21 secs | 00:06 |
benji_york | hmm, I'm adding profile support to the new test runner (we're using) so I can profile testbrowser :) | 00:07 |
srichter | ok | 00:08 |
J1m | I'm seeing the same test failure | 00:13 |
J1m | Someone fixed a spelling error. | 00:14 |
srichter | erich did check in a bunch of spelling errors | 00:15 |
J1m | fdrake, this is easy to fix, but someone needs to spank him. | 00:16 |
fdrake | so it was 38046 that broke it? | 00:16 |
fdrake | sounds like it from Benji's and my tests | 00:16 |
*** barsema has quit IRC | 00:17 | |
fdrake | i see it now; ok | 00:18 |
fdrake | i'll fix and send Egon a note | 00:18 |
srichter | probably the branch is affected too | 00:19 |
fdrake | likely; I'll check and fix there if so | 00:19 |
fdrake | argh; too many checkouts! | 00:25 |
fdrake | ok, all better now... :-) | 00:41 |
fdrake | srichter, I've not seen a 3.1.0c2 release yet. | 00:41 |
fdrake | did I miss it? | 00:42 |
srichter | nope ;-) | 00:42 |
srichter | people have lept fixing stuff | 00:42 |
fdrake | you're running the risk that I'll change the build process again :-) | 00:42 |
srichter | and I was really busy over the weekend, since my advisor managed to give me a lot of work :-( | 00:42 |
srichter | that's ok | 00:42 |
fdrake | it's not a small commit, if you decide you want it on the branch | 00:43 |
fdrake | i'd rather move it to the branch after 3.1.0 | 00:43 |
srichter | ok | 00:43 |
*** RaAtBRC|away has quit IRC | 00:43 | |
fdrake | but I might not be patient enough :-) | 00:44 |
srichter | hehe | 00:44 |
fdrake | it's all I can do to wait for the tests to run on the trunk :-) | 00:45 |
*** J1m has quit IRC | 00:49 | |
fdrake | ok, i can commit my changes now that Jim's not watching... | 00:49 |
fdrake | :-) | 00:49 |
benji_york | :) | 00:55 |
*** tvon has joined #zope3-dev | 01:01 | |
fdrake | Jim will have a merry Wednesday morning when he sees what's under the "svn up". :-) | 01:04 |
*** bradb-bbl is now known as bradb | 01:04 | |
fdrake | srichter, that's a huge build-process patch | 01:08 |
fdrake | you should release 3.1.0 before I port it to the branch | 01:09 |
fdrake | :-) | 01:09 |
*** fdrake has quit IRC | 01:27 | |
*** benji_york has quit IRC | 01:44 | |
yota | what is the date release for zope 3.1 ? | 01:59 |
hazmat | can z3.app be used without security proxies just by removing the publishing wrapper? | 02:17 |
srichter | hazmat: I think so | 03:01 |
srichter | hazmat: you need to ensure that removeSecurityProxy() also works opn plain objects | 03:02 |
srichter | yota: good question; I will do RC2 this week for sure | 03:02 |
srichter | there were just a lot of bug fixes recently | 03:02 |
yota | great :) | 03:12 |
srichter | yota: btw, what's your real name? | 03:18 |
*** alga has quit IRC | 03:18 | |
yota | I am not yet developer (maybe in a few times if i'm able to understand all Zope3 machinery :) for now I work only on translation. My name is Sebastien Douche | 03:25 |
srichter | ah, nice to meet you sebastien | 03:25 |
srichter | I saw that you were working on the french translations | 03:25 |
yota | yes, currently for plone & zope3 | 03:26 |
srichter | oh great, a prefessional translator! :-) | 03:26 |
yota | and surely for z3ecm | 03:26 |
yota | hum, I hope that I will be able to make another things :) | 03:27 |
srichter | me too ;-) | 03:27 |
yota | I want create french website and maybe write a book, for q3/q4 2006 | 03:31 |
srichter | cool! | 03:31 |
yota | yes, it is. Before, I must learn Zope3 ! | 03:33 |
srichter | yota: do you have a publisher already? | 03:39 |
yota | no, the book is a projet | 03:41 |
yota | project | 03:41 |
yota | step by step, learn zope3 in depth, participate, coding stuff, increase my english skill ... | 03:44 |
srichter | oh, you don't want to write in Fench? | 03:46 |
srichter | oh, you don't want to write in French? | 03:46 |
yota | yes in french, there are already 2 books in English ;) | 03:48 |
srichter | right :-) | 03:50 |
yota | but to become an "expert", it is necessary to take part in sprints, ml discussions and thus to speak English. My english is broken/ | 03:50 |
srichter | I think France is a big market, sicne they support OSS so prominently | 03:51 |
srichter | yota: are you coming to the Neckarsprint? | 03:51 |
yota | np, I'm not and I regret it. I would like take part of zope3 & z3ecm develpment | 03:53 |
srichter | it's not too late | 03:56 |
srichter | yota: you can still come | 03:57 |
srichter | maybe you can travel with Florent and Julien | 03:57 |
yota | not it is too early. My English level is still too low :) | 03:59 |
*** bradb is now known as bradb-away | 03:59 | |
yota | 3:00 AM, time to sleep | 03:59 |
yota | gn srichter :) | 03:59 |
*** yota has quit IRC | 03:59 | |
*** SureshZ has quit IRC | 04:06 | |
*** SureshZ has joined #zope3-dev | 04:09 | |
*** RaFromBRC has joined #zope3-dev | 04:56 | |
*** stub has joined #zope3-dev | 05:00 | |
*** RaFromBRC has quit IRC | 05:07 | |
*** SureshZ has left #zope3-dev | 05:43 | |
*** d2m has quit IRC | 05:57 | |
*** d2m_ has joined #zope3-dev | 05:58 | |
*** d2m_ is now known as d2m | 05:58 | |
bob2 | is it possible to have zope.app.debug.Debugger write changes to the ZODB? | 06:05 |
srichter | I dunno | 06:11 |
bob2 | heh | 06:12 |
bob2 | or is there another way to interactively poke at it? | 06:12 |
srichter | aehm, of course you can use the debugger | 06:13 |
srichter | as long as you commit the transaction | 06:13 |
bob2 | ah | 06:13 |
bob2 | I guess that's my question then | 06:13 |
*** MiUlEr has joined #zope3-dev | 06:38 | |
bob2 | so | 06:41 |
bob2 | I want to have a permission system such that some users can add some object to some containers | 06:42 |
bob2 | but not others | 06:42 |
bob2 | is this going to need a new security policy? | 06:42 |
srichter | nooooo | 06:48 |
bob2 | heh | 06:48 |
srichter | just give those principals that can add stuff to the container the permission to do so (local permission) | 06:49 |
bob2 | hm | 06:49 |
bob2 | I guess I can't see how to configure this outside of zcml | 06:49 |
srichter | its just a matter of setting up the security mappings correctly | 06:51 |
srichter | is done via some adapters | 06:51 |
bob2 | hm | 06:52 |
bob2 | can you point me at any examples | 06:52 |
srichter | just look what the view code of the grant view is doing | 06:52 |
bob2 | your and phillipp's books don't seem to cover this sort of thing | 06:52 |
srichter | well, schooltool has some security setup code in python | 06:52 |
bob2 | hmm | 06:52 |
srichter | because it is not needed this often | 06:52 |
srichter | though I demonstrate the grant view | 06:53 |
bob2 | do "normal" apps statically configure permissions in zcml? | 06:53 |
srichter | well, you declare the permissions in ZCML | 06:53 |
srichter | however, ST very much relies on local permissions | 06:53 |
bob2 | local permissions = permissions applied to a certain object? | 06:53 |
bob2 | (just want to clarify) | 06:53 |
srichter | yes | 06:58 |
bob2 | hm | 06:59 |
bob2 | thanks again :) | 06:59 |
bob2 | it doesn't sound like that uncommon a use case | 07:05 |
bob2 | surely in CMS's its common for only the marketing manager to be able to update /buyourstuff/ or whatever | 07:05 |
srichter | probably | 07:05 |
bob2 | hm, I mean, is this uncommon, not implemented yet, or am I looking at it wrong and thus missing the obvious solution? | 07:06 |
srichter | no, it is all implemented | 07:06 |
srichter | all there | 07:06 |
srichter | schooltool uses very fined grained security control | 07:07 |
*** MiUlEr has quit IRC | 07:14 | |
*** deo has quit IRC | 07:16 | |
*** deo has joined #zope3-dev | 07:30 | |
*** zagy has quit IRC | 08:14 | |
*** zagy has joined #zope3-dev | 08:17 | |
*** sashav has quit IRC | 08:31 | |
*** Theuni has joined #zope3-dev | 08:51 | |
*** sashav has joined #zope3-dev | 09:01 | |
*** j-w has joined #zope3-dev | 09:13 | |
*** sashav has quit IRC | 09:20 | |
*** tvon has quit IRC | 09:35 | |
*** sashav has joined #zope3-dev | 09:52 | |
bob2 | what's the best way to initialise startup services for a zope3 system? | 09:55 |
bob2 | e.g. setup a logger | 09:55 |
bob2 | just hack runzope? | 09:55 |
*** jhauser has joined #zope3-dev | 09:58 | |
*** tvon has joined #zope3-dev | 10:09 | |
*** MJ has quit IRC | 10:14 | |
bob2 | are events handled in the same thread that triggered them? | 10:22 |
Theuni | should be that we | 10:26 |
Theuni | thatway | 10:26 |
Theuni | argh | 10:26 |
Theuni | should be that way | 10:26 |
bob2 | hm, dang | 10:26 |
*** d2m has quit IRC | 10:39 | |
*** tanghus has quit IRC | 10:53 | |
*** shivanand has joined #zope3-dev | 10:59 | |
*** shivanand has quit IRC | 11:04 | |
*** MJ has joined #zope3-dev | 11:06 | |
*** tanghus has joined #zope3-dev | 11:12 | |
*** tol has joined #zope3-dev | 11:16 | |
*** tol has left #zope3-dev | 11:20 | |
*** tol has joined #zope3-dev | 11:22 | |
*** timte has joined #zope3-dev | 11:22 | |
*** vlado has joined #zope3-dev | 11:25 | |
*** tol has quit IRC | 11:27 | |
*** tarek_ has joined #zope3-dev | 11:27 | |
*** tanghus_ has joined #zope3-dev | 11:28 | |
*** tanghus has quit IRC | 11:30 | |
*** tanghus_ is now known as tanghus | 11:39 | |
*** vlado has quit IRC | 11:56 | |
*** kevc has joined #zope3-dev | 12:11 | |
andrew_m | can someone please tell me if that configuration for adapting a content object to ILockable looks correct?: | 12:11 |
andrew_m | <adapter | 12:11 |
andrew_m | for=".interfaces.ICEContent" | 12:11 |
andrew_m | provides="zope.app.locking.interfaces.ILockable" | 12:11 |
andrew_m | factory="zope.app.locking.adapter.LockingAdapterFactory" | 12:11 |
andrew_m | permission="ceng.EditContent" | 12:11 |
andrew_m | /> | 12:11 |
andrew_m | the LockingAdapterFactory doc says "This should be registered against zope.interface.Interface to provide adaptation to ILockable.".. not sure what that means | 12:13 |
*** sashav has quit IRC | 12:17 | |
*** MrTopf has joined #zope3-dev | 12:29 | |
*** stub has quit IRC | 12:33 | |
*** wiggy has quit IRC | 12:44 | |
*** hdima has joined #zope3-dev | 12:52 | |
*** drzoltron_ has joined #zope3-dev | 12:55 | |
* drzoltron_ says hi | 12:56 | |
*** projekt01 has joined #zope3-dev | 13:03 | |
*** mkerrin has joined #zope3-dev | 13:12 | |
*** sashav has joined #zope3-dev | 13:19 | |
*** philiKON has joined #zope3-dev | 13:26 | |
*** kakella has joined #zope3-dev | 13:48 | |
andrew_m | hi philiKON: do you happen to know the locking package of zope? | 13:56 |
*** yota has joined #zope3-dev | 14:04 | |
*** J1m has joined #zope3-dev | 14:11 | |
*** benji_york has joined #zope3-dev | 14:15 | |
andrew_m | phuu.. got locking working.. strongly suggest that module documentation should also contain hints about how to configure the module and what other special modules have to be installed to make it work (in this case zope.app.keyreference) | 14:19 |
benji_york | andrew_m, I guess the locking package didn't have zpackage metadata | 14:20 |
J1m | By "installed", I assume he meant configured, | 14:21 |
benji_york | ahh, ok | 14:21 |
*** stub has joined #zope3-dev | 14:21 | |
andrew_m | well.. had to put zope.app.keyreference-configure.zcml into package-includes | 14:21 |
benji_york | that means zope 3 needs layered tests | 14:21 |
srichter | I want to try a new approach to evolutions | 14:22 |
J1m | In this case, better documentation would probably sufficed. | 14:22 |
srichter | ST does not have to worry about API BBB at this point | 14:22 |
J1m | good documentation isn't easy. | 14:22 |
srichter | only data compat | 14:22 |
J1m | ST? | 14:23 |
srichter | I would really like to hook into the broken object support and look up the new class via a mapping | 14:23 |
srichter | do you think this is a good approach? | 14:23 |
srichter | ST = SchoolTool | 14:23 |
J1m | ah | 14:23 |
benji_york | J1m, yeah, that wouldn't hurt, but I'm a fan of executable documentation :) | 14:24 |
srichter | I tried an evolution script that converts broken objects to classes | 14:24 |
J1m | It would be nice to come up with better data migration strategies for ZODB. | 14:24 |
srichter | however, we have circular references and a simple recurrence model does not help | 14:24 |
J1m | benji_york, well, in this case, the setup code would probably have given some insights and probably should have been presented in the documentation. | 14:26 |
andrew_m | benji_york: the docu was runnable, but i had to 'map' the test code into configuration which harder than having an example configuration in the doc | 14:26 |
andrew_m | is harder | 14:26 |
andrew_m | also some things in the test are done by zope, some things i have to do | 14:26 |
J1m | srichter, generally, for evolution, I would want to have the old classes around until the evolution was complete. | 14:27 |
benji_york | J1m and andrew_m, right, I was saying that if it had been zcml to start with, it would have been more helpful | 14:27 |
benji_york | no biggie, just an observation | 14:27 |
andrew_m | benji_york: yep | 14:27 |
J1m | It sounds like, in your case, you really want to convert object graphs, rather than individual objects. | 14:27 |
srichter | yeah, this is just really painful in this case, since I got rid of an entire package, and I would have to at least simulate until the evolution is done | 14:27 |
J1m | Can't you get the package back? | 14:28 |
srichter | well, I want to do this as a last resort | 14:28 |
srichter | I did only rearraning, so I could simply have an empty shell that imports the classes from their new location | 14:30 |
srichter | I really just want to specify a mapping of old_class_path -> new_class_path and the ZODB should do the right thing | 14:32 |
*** alga has joined #zope3-dev | 14:32 | |
*** ignas has joined #zope3-dev | 14:32 | |
J1m | srichter, the usual way to do this is via aliasing | 14:32 |
srichter | but aliasing works only for modules, right? | 14:33 |
J1m | I thought you were doing more than moving classes around | 14:33 |
srichter | but I split packages | 14:33 |
srichter | so there was a lot of stuff in st.app, not it is in st.person.person, st.group.group, ... | 14:33 |
srichter | so there was a lot of stuff in st.app, now it is in st.person.person, st.group.group, ... | 14:34 |
srichter | can module aliasing do this? | 14:34 |
J1m | Give me an example of a changed class path, | 14:35 |
srichter | ok | 14:35 |
srichter | schooltool.app.PersonContainer --> schooltool.person.person.PersonContainer | 14:35 |
srichter | schooltool.app.Person --> schooltool.person.person.Person | 14:35 |
J1m | and schooltool.app is gone? | 14:35 |
srichter | no, this one is not, but: | 14:36 |
*** drzoltron_ has quit IRC | 14:36 | |
srichter | schoolbell.app.app.SomeObject --> schooltool.some.some.SomeObject | 14:36 |
srichter | (there are several SomeObject cases, not just one) | 14:36 |
J1m | so schooltool.app.app is gone? | 14:37 |
srichter | no | 14:37 |
J1m | ok. so anyway, | 14:37 |
srichter | actually schooltool.app was a module and is now a package | 14:37 |
J1m | you could import these classes into schooltool.app. | 14:38 |
J1m | Or. you could poke them there. | 14:38 |
srichter | right | 14:38 |
srichter | what sbout the schoolbell case? | 14:38 |
J1m | btw, your module depreication code is rather expensive. | 14:39 |
srichter | :-( | 14:39 |
srichter | we can look at it next week | 14:39 |
J1m | k | 14:39 |
J1m | so is schoolbell.app gone? | 14:39 |
J1m | so is schoolbell.app.app gone? | 14:39 |
srichter | yeah | 14:39 |
J1m | ok, so you create a fake schoolbell.app module and poke/import the classes there. | 14:40 |
srichter | I am thinking about hooking into ZODB.serialize.ObjectReader.getClassName; if I can insert my mapping there, I am fine | 14:41 |
srichter | right, though I think this is sub-optimal | 14:41 |
srichter | it makes all this much harder than it should be | 14:41 |
J1m | Perhaps. | 14:41 |
J1m | I'd rather you not hack ZODB. | 14:41 |
J1m | Feel free to propose a mechanism. | 14:41 |
srichter | ok | 14:42 |
srichter | will Tim be in F12g next week? | 14:42 |
srichter | maybe I can discuss some ideas with him | 14:42 |
srichter | or you for that matter :-) | 14:42 |
J1m | But, I suspect that introducting some ZODB registry is no better than using sys.modules. | 14:42 |
J1m | It would be nice to include Tim. | 14:42 |
J1m | I'll try to arrange some time with him. | 14:43 |
srichter | great | 14:43 |
J1m | The party might be the best time though, assuming that he's coming. | 14:43 |
srichter | I am playing around with it this week anyways, so I will have some experience by then | 14:43 |
srichter | ok | 14:43 |
srichter | J1m: oh, we need to come up with some topics for next week | 14:44 |
J1m | widget/form cleanup is one. | 14:44 |
J1m | I think Gary might write up a proposal. | 14:45 |
srichter | I originally thought about developing proposals for some of the upcoming work, but we can also code (I don't care) | 14:45 |
J1m | I think Dylan is interested in working in that area. | 14:45 |
srichter | cool | 14:45 |
srichter | I definitely want to talk to you and Gary about the twisted integration and the server startup code cleanup that will come with it. I have the use cases which I should write down until next week | 14:46 |
J1m | Both Gary and benji_york will participate. Perhaps Fred. | 14:47 |
benji_york | Jim, I just wish Benji and fdrake were going to be there <wink> | 14:50 |
*** alga has quit IRC | 14:53 | |
*** alga has joined #zope3-dev | 14:56 | |
*** regebro has joined #zope3-dev | 14:58 | |
MJ | Durn | 14:58 |
MJ | Just when something interesting is possible with a GMail account I find that mine has expired. | 14:59 |
benji_york | MJ, talk.google.com? | 14:59 |
MJ | Should've used it once in while.. | 14:59 |
MJ | benji_york: Yes | 14:59 |
MJ | I want to see what extra services they added. | 14:59 |
srichter | MJ: I am checking in KMail every so often | 14:59 |
MJ | Yeah, should have done so too | 15:00 |
MJ | So, are invites still thick on the ground these days or am I out of luck? | 15:00 |
srichter | Email> | 15:01 |
srichter | Email? | 15:01 |
MJ | srichter: I am shocked.. | 15:01 |
MJ | You don't know it? ;) | 15:01 |
MJ | mj@zopatista.com | 15:02 |
MJ | or whatever you fancy before the @ | 15:02 |
srichter | MJ sent | 15:02 |
MJ | Thanks | 15:02 |
MJ | Darn | 15:03 |
MJ | Why don't they delete useful gmail names when expired.. | 15:03 |
MJ | Well I'll be darned | 15:05 |
MJ | Now my ole account is active again. | 15:05 |
MJ | Must be the beta part in action. | 15:05 |
MJ | Thanks anyway Stephan, it turns out I can still use the old account. | 15:06 |
MJ | 50 invites to go.. | 15:06 |
*** bskahan has joined #zope3-dev | 15:07 | |
*** MJ has quit IRC | 15:18 | |
*** J1m has quit IRC | 15:26 | |
*** J1m has joined #zope3-dev | 15:26 | |
*** alga has quit IRC | 15:36 | |
*** tvon has quit IRC | 15:41 | |
*** mgedmin has joined #zope3-dev | 15:44 | |
*** alga has joined #zope3-dev | 15:50 | |
*** faassen has joined #zope3-dev | 16:02 | |
*** vlado has joined #zope3-dev | 16:05 | |
*** projekt01 has quit IRC | 16:23 | |
*** bradb-away is now known as bradb | 16:28 | |
*** fdrake has joined #zope3-dev | 16:44 | |
fdrake | has anyone tested the updated zpkg-based build on Windows? | 16:46 |
hdima | fdrake: seems like functional tests is broken now for svn checkouts.. | 16:47 |
fdrake | ;-( | 16:47 |
fdrake | just for Windows? | 16:47 |
hdima | since 'zopeskel/etc/package-includes/' is empty | 16:47 |
hdima | no | 16:48 |
* hdima use Linux | 16:48 | |
fdrake | do you build using make? | 16:48 |
hdima | no | 16:48 |
fdrake | or python setup.py build_ext -i ? | 16:48 |
fdrake | use make :-) | 16:48 |
J1m | probably need to tell people they need to run make in existing checkouts | 16:48 |
hdima | Oh, seems like I need to rebuild all... :-) | 16:48 |
fdrake | the setup.py command should be: | 16:48 |
fdrake | python setup.py build_ext -i install_data --install-dir . | 16:49 |
fdrake | yeah; i'll send a note to zope3-dev | 16:49 |
hdima | It's works now, thanks | 16:54 |
fdrake | np | 16:56 |
J1m | fdrake, it would be good to create a short page on how to setup zpkg info for simple Python packages. | 17:01 |
fdrake | yes, will do | 17:02 |
J1m | This should be aimed at package developers who don't need to actually use the zpkg tools. | 17:02 |
fdrake | right; basically just DEPENDENCIES.cfg and SETUP.cfg, right? | 17:02 |
J1m | yes, and perhaps menion the other one. | 17:03 |
J1m | (that has release meta data) | 17:03 |
J1m | yes, and perhaps mention the other one. | 17:03 |
fdrake | hmm, yeah, PACKAGE.cfg is somewhat useful, though unusual for simple packages | 17:03 |
fdrake | and PUBLICATION.cfg is only useful for something that's directly releasable (for someone using zpkg directly) | 17:04 |
J1m | true | 17:04 |
*** hdima has quit IRC | 17:08 | |
*** kakella has quit IRC | 17:11 | |
*** gbastien has joined #zope3-dev | 17:19 | |
*** gbastien is now known as __gotcha | 17:19 | |
__gotcha | srichter: hi | 17:20 |
srichter | hi | 17:20 |
__gotcha | srichter: have you heard of effort to pot apidoc to Five ? | 17:20 |
srichter | no | 17:20 |
__gotcha | would that be easy ? does apidoc depends a lot on zope 3 ? | 17:20 |
srichter | apidoc is 100% Zope 3 specific | 17:21 |
__gotcha | sure | 17:21 |
srichter | and I mean in the way the data is collected | 17:21 |
__gotcha | would be nice though to be able to use apidoc for the five part of 2.8 | 17:21 |
benji_york | srichter, I just added an optimization to ClientForm that sped up a test run of zope.testbrowser's README.txt from (a three-run average) of 6.96 seconds to 6.23 seconds | 17:21 |
srichter | it knows very specifically how to tickle information out of view factories, etc | 17:21 |
srichter | benji_york: oh, cool; | 17:22 |
__gotcha | srichter: iow, very few chance to port it to Five without a lot of work | 17:22 |
srichter | __gotcha: well, you would have to write all the inspection code yourself | 17:22 |
srichter | __gotcha: Would you want this port to work on Zope 2 objects as well? | 17:23 |
__gotcha | first on five objects | 17:23 |
__gotcha | lets be humble and realistic | 17:23 |
srichter | If they are similar enough to the usual Zope 3 way of registering and reuse those directives, it is possible | 17:24 |
srichter | I would just try it | 17:24 |
__gotcha | I just tried to inluce apidoc configure.zcml and it chokes on the class directive | 17:24 |
__gotcha | Ill have to look deeper later | 17:24 |
__gotcha | Ill contact you back after reading apidoc code | 17:24 |
__gotcha | thanks | 17:25 |
srichter | ok | 17:25 |
*** __gotcha has quit IRC | 17:25 | |
benji_york | J1m, I got profiling working in the new test runner, including combining each subprocess' data into the final report | 17:30 |
J1m | cool | 17:32 |
mgedmin | "the new test runner"? where is it? | 17:33 |
J1m | I assume you figured out how to combine pstat files. | 17:33 |
J1m | mgedmin, http://svn.zope.org/zope.testing/trunk/src/zope/testing/testrunner.py?rev=38027&view=log | 17:34 |
benji_york | J1m, nope, the pstat.Stat objects have a method that will combine the data from one object to another, so I just generate a file for each subjprocess and then combine them together at the end | 17:34 |
J1m | isn't that what I said? | 17:34 |
benji_york | there is a problem though, I hit a bug in 2.4.1 with the profiler (already fixed in CVS) that, strangely, means you can only profile the tests when using the -O option to Python | 17:34 |
benji_york | J1m, sure :) | 17:35 |
J1m | mgedmin, perhaps I'll switch the z3 trunk to use it this week. | 17:35 |
J1m | benji_york, interesting | 17:35 |
J1m | Good thing we don't require 2.4 ;) | 17:35 |
srichter | :-\ | 17:36 |
benji_york | J1m, using the profiler I've already made an improvement to (Gary's pending version of) ClientForm that sped up zc.task tests from 34 seconds to 29 | 17:36 |
J1m | cool | 17:36 |
benji_york | J1m, I *believe* the bug also exists in 2.3, I'll have to double-check though | 17:36 |
J1m | I suppose this means that you could profile the tests only. | 17:37 |
J1m | (without having to profile the test runner at the same time) | 17:37 |
benji_york | J1m, nope, I haven't put any effort into isolating the test code, but I don't think it would be too hard to do that | 17:38 |
benji_york | J1m, the bug is also in 2.3.4 | 17:38 |
J1m | Must be fixed in 2.3.5 :) | 17:39 |
benji_york | nope | 17:40 |
benji_york | or, maybe :) | 17:40 |
benji_york | nope, can't be in 2.3.5 because the bug wasn't discovered and fixed until after 2.3.5 shipped | 17:41 |
benji_york | zope 3.1 should require 2.4.2 <evil grin> | 17:42 |
fdrake | that works for me :-) | 17:45 |
stub | J1m: Would you object to the new test runner supporting something similar to my resources? A resource as I think of it would be identical to the existing layers, so the change would be a test could specify multiple layers instead of just one. | 17:49 |
stub | erm... in english that means 'Make the new test runner support tests that specify multiple layers instead of just one' | 17:50 |
J1m | I think that would be fine, especially if you implemented it including tests. :) | 17:52 |
benji_york | J1m, because of the requirement to run in optimized mode for the profiler, I can't think of a good way to do tests for the profiler code... | 17:56 |
benji_york | so should I just sit on the changes? | 17:56 |
*** MrTopf has quit IRC | 17:56 | |
J1m | Hm, what is the issue with the profiler and -O? | 17:57 |
J1m | I use the profiler all the time without -O. | 17:57 |
benji_york | it's complicated, I can explain if you want, but the bottom line is that you can't run bin/test --profile unless you also run Python with -O | 17:58 |
mgedmin | strange | 17:59 |
*** sashav has quit IRC | 18:01 | |
stub | J1m: ok. The other issue that got brought up in discussions with my cow orkers was that the seperation of powers seems wrong. One idea seems to be that the testrunner should assemble the list of tests, stuffing them in a test suite and then calls testsuite.run(). The test cases inherit from a common base class which contains the intelligence to setup the layers if they are not already setup, and to tear down the layers if they are not needed | 18:04 |
stub | J1m: The advantage being that you could use any test runner, such as a GUI, and have the tests still use all the correct layer fixtures | 18:05 |
stub | But I think that is a longer term thing, and you might not agree. There are extra tricks needed to keep functionality, such as the TestSuite needing to fork in order to restart the tests in a new process. | 18:06 |
stub | I also don't know if I personally can be bothered for Launchpad, but I think there are rumblings for a better test environment for the Bazaar project and possibly some other bits and pieces. | 18:07 |
mgedmin | this might not be exactly relevant, but recently we started experimenting with a distributed test runner (rsync source into N machines, use ssh + python on the other end to run the test runner in slave mode; report results as they come locally) | 18:08 |
mgedmin | the current prototype mostly works and is based on SchoolTool's test runner, but I'd like to base in on the Z3 one | 18:08 |
mgedmin | for that I need to asking the test runner to run tests one by one | 18:09 |
*** d2m has joined #zope3-dev | 18:10 | |
stub | If TestSuite.run yielded results as they came in, that would work (and again, you could plug in a different front end instead of TextTestRunner or whatever) | 18:11 |
stub | (if that is what you were referring too) | 18:12 |
ignas | stub, and if you want to run the test1 from the test suite on slave1 and test2 + test3 on slave2 ... | 18:12 |
mgedmin | stub, currently we create a bunch of TestSuites that contain one test each, then pick the results form a custom TestResult class | 18:14 |
stub | ignas: What my co workers were suggesting is pretty much to just move that intelligence from the test runner into the test suite, which makes the front end pluggable. | 18:14 |
* stub isn't fussed either way personally, since I've never wanted to use a GUI testrunner | 18:14 | |
fdrake | Basic information on the build/packaging support files is documented at http://www.zope.org/Wikis/DevSite/Projects/ComponentArchitecture/BuildAndPackagingInfo | 18:23 |
fdrake | I think that's link from a reasonable place (http://www.zope.org/Wikis/DevSite/Projects/ComponentArchitecture/Zope3DeveloperInfo) | 18:24 |
* J1m gets off phone w Benji | 18:34 | |
*** regebro has quit IRC | 18:34 | |
J1m | stub, another use case is to be able to select tests by layer. | 18:37 |
J1m | layers might also be a nice basis for distribution of processing. | 18:38 |
J1m | Run layer 1 on machine 1, layer 2 on machine 2, and so on. | 18:38 |
J1m | Of course, if you have a layer per test ... :) | 18:39 |
stub | J1m: Ok. That would involve passing more information into the TestSuite, which will annoy the xUnit nazis who were pushing this ;) | 18:39 |
stub | J1m: Nah - I'll just have a mix of about 1-7 layers per test ;) | 18:39 |
J1m | But the test runner already has logic for combining results from payers when they are run in separate subprocesses. | 18:40 |
J1m | right, I view each mix as a layer. | 18:40 |
J1m | In fact, if I was to implement support for multiple layers, I *might* implement it by actually defining layers on the fly. | 18:41 |
stub | Yes - that was how I was thinking of doing it. It could have been done using a helper that generates the class on the fly, except that we then couldn't run all the tests that 'mix-in' the librarian layer for example. | 18:43 |
stub | ie. class Foo(TestCase): layer = build_layer(Librian, Db, XmlRPC) | 18:43 |
J1m | right | 18:48 |
stub | so list of layers would be better, and the test runner would know that the test doesn't care in which order the layers are setup and can optimize the order appropriately, but in the above spelling the test runner couldn't tell if it was a mix of layers or a 'traditional' layer. | 18:57 |
J1m | right | 19:00 |
*** stub has quit IRC | 19:07 | |
*** bradb has quit IRC | 19:11 | |
*** bradb has joined #zope3-dev | 19:17 | |
*** zagy has quit IRC | 19:27 | |
*** sashav has joined #zope3-dev | 19:28 | |
*** philiKON has quit IRC | 19:33 | |
*** vlado has quit IRC | 19:38 | |
*** bradb is now known as bradb-lunch | 19:40 | |
*** segoave has joined #zope3-dev | 19:42 | |
*** tarek_ has quit IRC | 19:43 | |
*** J1m has quit IRC | 20:01 | |
*** j-w has quit IRC | 20:10 | |
*** RaFromBRC has joined #zope3-dev | 20:22 | |
*** RaFromBRC is now known as RaFromBRC|busy | 20:23 | |
*** RaFromBRC|busy has quit IRC | 20:31 | |
*** bradb-lunch is now known as bradb | 20:35 | |
fdrake | srichter, what's the schedule for 3.1.0c2? | 20:35 |
srichter | this week :-) | 20:35 |
fdrake | my weeks end on Wednesdays, how about yours? :-) | 20:36 |
srichter | Sunday :-) | 20:37 |
fdrake | sheesh, your week is busted :-) | 20:37 |
srichter | (remember that in my books the last day of the week is Sunday -- I am from Europe) | 20:37 |
fdrake | and in my books, I want 3.1.0 out, so I can merge all the build changes in without touching 3.1.0 | 20:38 |
srichter | yep, I know | 20:39 |
srichter | I don't mind if someone else starts the process | 20:39 |
srichter | I am just extremely busy, but I am getting done | 20:40 |
fdrake | I don't even remember the process. | 20:40 |
srichter | :-) | 20:40 |
fdrake | And no, I don't read wikis. :-) | 20:40 |
srichter | I just follow MakeARelease all the time :-) | 20:41 |
fdrake | Somehow, I suspect that's a wiki page. Since I wrote it, I'm incapable of reading it. | 20:42 |
fdrake | (wrote it initially, that is) | 20:42 |
srichter | LOL | 20:43 |
srichter | no, I initally wrote it, but you worked on it alot | 20:43 |
fdrake | I'm supposed to be trying to get the Zope 2 trunk to use a new-style zpkg-based build. | 20:44 |
fdrake | It's been a long time since I've looked at that. :-( | 20:44 |
*** RaFromBRC has joined #zope3-dev | 20:55 | |
fdrake | i guess there's still a zope-dev for Zope 2... | 20:56 |
*** faassen has quit IRC | 20:56 | |
fdrake | guess not... | 20:56 |
*** Theuni has quit IRC | 20:57 | |
srichter | zope-dev for Zope 2? | 20:57 |
fdrake | right...., looks like there's just #zope | 20:58 |
fdrake | I'll get over it. ;-) | 20:59 |
srichter | oh, I see | 20:59 |
srichter | if you feel better I join #zope-dev, though I have no clue :-) | 21:00 |
fdrake | nah, don't bother, but thanks for the offer ;-) | 21:00 |
fdrake | i'm more interested in whether anyone already there has a clue | 21:01 |
fdrake | or if there are still even people attached to those nicks | 21:02 |
srichter | he he | 21:03 |
fdrake | feels like a graveyard over there.... | 21:03 |
srichter | he he | 21:06 |
fdrake | maybe all the remaining Zope 2 users are in Europe, and safely tucked into bed.... | 21:07 |
srichter | you know this is probably true | 21:17 |
fdrake | makes me wonder what happened to all the Aussies | 21:19 |
*** RaFromBRC has quit IRC | 21:21 | |
*** RaFromBRC has joined #zope3-dev | 21:22 | |
*** AJC has joined #zope3-dev | 21:28 | |
*** bradb has quit IRC | 21:34 | |
*** lunatik has joined #zope3-dev | 21:35 | |
*** lunatik has left #zope3-dev | 21:36 | |
*** bradb has joined #zope3-dev | 21:36 | |
*** mkerrin has quit IRC | 21:38 | |
*** bradb has quit IRC | 21:49 | |
*** bradb has joined #zope3-dev | 21:54 | |
*** RaFromBRC is now known as RaFromBRC|away | 22:06 | |
*** timte has quit IRC | 22:19 | |
*** tvon has joined #zope3-dev | 22:25 | |
*** ignas has quit IRC | 22:32 | |
*** lunatik has joined #zope3-dev | 22:35 | |
*** MJ has joined #zope3-dev | 22:46 | |
*** donkey has joined #zope3-dev | 22:51 | |
*** AJC has quit IRC | 22:52 | |
*** J1m has joined #zope3-dev | 23:06 | |
*** bradb has quit IRC | 23:12 | |
*** AJC has joined #zope3-dev | 23:17 | |
*** bradb has joined #zope3-dev | 23:17 | |
*** donkey has quit IRC | 23:18 | |
*** RaFromBRC|away is now known as RaAtBRC | 23:21 | |
*** lunatik has left #zope3-dev | 23:27 | |
*** yota has quit IRC | 23:28 | |
*** yota has joined #zope3-dev | 23:29 | |
*** bskahan has quit IRC | 23:29 | |
benji_york | srichter, ayt? | 23:48 |
srichter | yeah | 23:48 |
srichter | benji_york: pong | 23:49 |
benji_york | when you get a chance try out the latest testbrowser on your tests that went from 8 seconds to 21, I've made the tests (that I've been profiling) run about twice as fast | 23:49 |
srichter | oh, awesome | 23:49 |
srichter | let me make a checkin and then update my testbrowser version | 23:50 |
benji_york | cool | 23:50 |
*** donkey has joined #zope3-dev | 23:53 | |
*** AJC has quit IRC | 23:54 | |
*** alga has quit IRC | 23:56 | |
*** mgedmin has quit IRC | 23:56 | |
*** AJC has joined #zope3-dev | 23:57 | |
*** jhauser has quit IRC | 23:57 | |
*** donkey has quit IRC | 23:58 | |
srichter | benji_york: it is quiet abit fast; from 22 to 17 secs | 23:58 |
srichter | so 25% increase | 23:58 |
srichter | pretty good | 23:58 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!