IRC logs of #tryton for Monday, 2008-12-08 #tryton log beginning Mon Dec 8 00:00:01 CET 2008
-!- yangoon( has joined #tryton05:18
-!- udono( has joined #tryton06:32
-!- Timitos(n=Timitos@ has joined #tryton07:06
-!- sharkcz( has joined #tryton07:39
-!- Cristi_an(n=Cristi@ has joined #tryton08:45
-!- cedk(n=ced@gentoo/developer/cedk) has joined #tryton08:48
-!- nicoe( has joined #tryton09:24
-!- cedric_b( has joined #tryton09:29
-!- Gedd( has joined #tryton09:47
-!- ecir3( has joined #tryton09:57
CIA-54tryton: C?dric Krier <> default * 1093:ce21962c2ed9 tryton/tryton/common/ ( Use the right param for properties10:29
CIA-54tryton: C?dric Krier <> default * 1094:0a24b798fcfb tryton/tryton/gui/window/ (3 files in 2 dirs): Force color of title10:29
CIA-54tryton: C?dric Krier <> default * 1095:d21a1956663f tryton/tryton/ (5 files in 3 dirs): Improve colors management to use the default style colors as much as possible10:29
cedkX0d_of_N0d: hi10:39
cedkX0d_of_N0d: I think you try to use a dark theme with the client10:40
cedkX0d_of_N0d: with the changeset d21a1956663f, it must look better10:40
-!- Cristi_an__(n=Cristi@ has joined #tryton10:47
-!- bechamel( has joined #tryton11:14
cedkudono: one more time, a great idea taken from Tryton :-)11:54
cedkudono: we will see the quality of the documentation11:55
cedkit will be perhaps the wiki converted into ReST11:58
udonocedk: or they put the english handbook into rst...12:24
udonocedk: but anyway I find 650 pages too much...12:25
CIA-54tryton: matb roundup * #649/Unable to create new employee: [new] Error on creating new employee: The field "Account Payable" on "Party" is required.14:00
CIA-54tryton: C?dric Krier <> default * 176:3b8b28561fbf purchase/ Add docstring to create_invoice14:09
cedkudono: I think it must be the wiki:
cedkudono: there is about 750 pages on the wiki14:49
-!- X0d_of_N0d(i=C-C_C-X@gateway/tor/x-d4d06cfbfdf95fea) has joined #tryton15:21
CIA-54tryton: C?dric Krier <> default * 1096:e6e77a18bfcb tryton/TODO: Add todo for pysocket and **kwargs15:55
CIA-54tryton: C?dric Krier <> default * 1097:8c9b4bfad47d tryton/ (COPYRIGHT tryton/gui/window/
CIA-54tryton: Add sednacom to contributor and remove Async Open Source from copyright as15:55
CIA-54tryton: date_time was rewritten15:55
yangooncedk: tryton -h tells option   -d LOG_LOGGER, --log=LOG_LOGGER16:03
yangooncedk: this switch is not in the docs16:03
yangooncedk:  I currently want to debug/log for more information for the network errors I am getting constantly on long tasks16:04
cedkyangoon: it is just a timeout16:05
yangooncedk: timeout which is configurable?16:05
yangooncedk: because the problem is i.e., that after timeout other wizards are not running any more16:06
yangooncedk my db server is surely not that fast, but it should be usable without problems nevertheless16:07
cedkyangoon: the timeout is fixed to 12016:08
cedkyangoon: but I'm thinking about a solution to increase it16:08
yangooncedk: ok, thx16:09
yangooncedk what about option -d?16:09
yangooncedk: shouldn't it be documented?16:09
cedkyangoon: I don't know if it works16:09
cedkTimitos: hi16:31
Timitoscedk: hi16:32
-!- nicoe( has joined #tryton16:55
CIA-54tryton: matb roundup * #628/Client: network error on heavy load: It seems, that this network error is caused by some timeout on long tasks. I get it regularly on creation of the german account chart. The networ ...17:11
CIA-54tryton: matb roundup * #648/UI alignament BUG: Confirming that this seems indeed to be due to the gtk2 theme. And just the default theme 'Raleigh' (at least on Debian testing with KDE here) is ...17:15
CIA-54tryton: Bertrand Chenal <> default * 213:b3b12935b55e website/sphinx_data/templates/layout.html: fix sphinx template for xhtml validation17:19
CIA-54tryton: Bertrand Chenal <> default * 214:54ce875341b1 website/: merge17:19
-!- mmarshall( has joined #tryton17:46
cedkmmarshall: hi17:52
CIA-54tryton: C?dric Krier <> default * 1098:f7210b6ff9f4 tryton/tryton/gui/window/ Add email address for Sednacom18:20
cedkmmarshall: if you have any question about Tryton, you can ask here18:33
CIA-54tryton: C?dric Krier <> default * 1099:9110598f7b1f tryton/tryton/ Use small timeout to connect and after increase it to 1 hour for issue62818:40
CIA-54tryton: ced roundup * #628/Client: network error on heavy load: [resolved] Fix with changeset 9110598f7b1f18:41
CIA-54tryton: C?dric Krier <> default * 1332:0a5d9fa9c3e1 trytond/trytond/ Use small timeout to connect and after increase it to 1 hour18:42
-!- Cristi_an(n=Cristi@ has joined #tryton18:44
mmarshallcedk: I'm a little curious as to why there are so many ex-tiny employees working on CRM/ERP solutions. :-)18:47
mmarshallI see b2ck and openhex... are there others?18:47
cedkmmarshall: I don't think18:48
mmarshallI have question about model inheritance:  when you instance an OSV subclass with a previously used _name and add a new field, is the new field added to the original db table or is a new table created behind the scenes with the new field and a foreign key reference to the old table?19:24
-!- Cristi_an__(n=Cristi@ has joined #tryton19:28
mmarshallLooking at the "Models Inheritance" example in the docs... it appears that what is happening isn't 'inheriting', but 'extending' (or 'monkey patching' as some one say.)19:28
mmarshallThe new Party class doesn't create a new model, but extends the original model.19:28
mmarshallIs this correct?19:28
CIA-54tryton: matb roundup * #650/Items reappearing untranslated in new database: [new] another set of items, that were already translated and exported, but appear again as untranslated: "Mark for Uninstallation (beta)" "" "Fal ...19:30
CIA-54tryton: matb roundup * #628/Client: network error on heavy load: [chatting] works great, thx!19:31
CIA-54tryton: ced roundup * #650/Items reappearing untranslated in new database: [testing] You must delete the English version of those records and after run an update all.19:38
cedkmmarshall: it is a mix of the both19:38
cedkmmarshall: as the method are called like a inheritage19:39
cedkmmarshall: you can see many place where we call it with super()19:39
cedkmmarshall: but you are partially right that it is an extend19:39
cedkmmarshall: because the previous version on the model is replaced in the pool by the new one that inherit from it19:40
cedkso it is a real inheritage (like Python) but we use only the latest instance of a model19:41
bechamelmmarshall: you question was about python objects or about the db tables ?19:41
mmarshallI see20:00
mmarshallI was asking about the data modeling (db tables)20:01
mmarshallBut I see that there isn't a direct relationship between the python classes and db tables.20:02
mmarshallBack to the models inheritance example, Party.__init__ is calling "super(Party, self).__init__()".  It looks that that just calls OSV.__init__, but is there some trick going on behind the scenes so that the original party class's __init__ is called?20:06
mmarshallAre metaclasses used so that the new Party class is switched from inheriting directly from OSV to be inheriting from the old Party class?20:07
bechamelmmarshall: the original __init__ was already called because each class is instanciated just after his definition20:08
mmarshallOh, duh.20:09
mmarshallI'm used to how django and elixir do it :)20:09
bechamelmmarshall: the behind the scene trick is thaht osv check if alreayd exist in the poll of osv classes  and add "manualy" the fields/function/constrains  on the original class20:10
bechamelmmarshall: you know elixir ? i understand your question about ex-tiny employee now :)20:11
mmarshallYeah, finding different ex-tiny employees working on erp products seemed kind of weird :D20:14
cedkmmarshall: it is a realy Python inheritance, it doesn't add "manually" function, etc...20:22
cedkthat is why super() works20:22
mmarshallSo, in the documentation there are some examples using fields.Function and a 'get_total' method.  Could a new class override get_total and call super(NewClass, self).get_total(...) and have it work as expected?20:24
mmarshallAnd would self.__class__.__bases__ give the previous class for that model instead of OSV?20:26
mmarshallMaybe I need to read the source :)20:27
-!- Cristi__(n=Cristi@ has joined #tryton20:27
bechamelmmarshall: for the super, yes: you can do res = super(NewClass, .. and later res[i] = other_value and then return res20:29
cedkbechamel: look at trytond/osv/ line 177, the function create_instance20:34
cedkbechamel: oups20:34
cedkmmarshall: ping20:34
-!- LordVan(n=lordvan@gentoo/developer/LordVan) has joined #tryton20:45
LordVancedk, poke20:46
LordVanbug in ebuilds .. openoffice-interact has missing space in dependency  '|| (app-office ..' shoudl be '|| ( app-office ...'20:50
mmarshallcedk: Ah, I see now: yet another class is created that subclasses from both the new class and the previous class.20:52
LordVanalso can't download openoffice-interact20:52
cedkLordVan: I see, the website doesn't keep older version21:01
cedkLordVan: and use a strange version numbering21:01
LordVanmaybe mirror a known2work version?21:02
CIA-54tryton: C?dric Krier <> default * 137:87108071048e gentoo-overlay/dev-python/openoffice-interact/ (Manifest openoffice-interact-0.1.20080602.ebuild): Fix typo and change download url21:05
cedkLordVan: done21:05
cedkmmarshall: yes, and we keep a disctionnary (POOL) with the latest class per _name21:08
-!- lordvan_(n=lordvan@gentoo/developer/LordVan) has joined #tryton21:11
-!- Gedd(n=ged@ has joined #tryton21:12
lordvan_duno what u last got cuz i had a disconnect, but i sync'd the ebuilds successfully merging now ;021:12
lordvanbtw :D21:30
X0d_of_N0dcedk: your change seems to have fixed the problem, but the hg repo seems a little unstable right now, is that to be expected?21:32
bechamellordvan: yes they are listed here :)21:32
cedkX0d_of_N0d: the version 1.1 yes, but the 1.0 no21:33
X0d_of_N0dI was pulling from 1.1, so yeah, I got it to work well enough, but it was kind of messy21:33
X0d_of_N0dbut your patch did fix the problem21:34
lordvanah ok :)21:34
cedkX0d_of_N0d: we try to keep the 1.1 working, but error can happen21:34
lordvancedk, current trytond ebuild doesn't install a trytond.conf in /etc21:35
CIA-54tryton: C?dric Krier <> default * 1053:bf90d56b3943 tryton/tryton/gui/window/view_form/view/form_gtk/ Fix completion on many2one for issue64621:35
CIA-54tryton: C?dric Krier <> default * 1054:9eb8efb24070 tryton/tryton/ (gui/ plugins/ Move plugin into plugins directory to be included in the sdist for issue64521:35
X0d_of_N0dI got a bunch of unmet dependances... "ERROR:init:module:relationship:Module not found!" a lot. but yeah, development branch it makes sense21:35
bechamelX0d_of_N0d: weird relationship doesnt exist anymore21:36
lordvangotta go cyas21:37
cedkX0d_of_N0d: you must have an old repo21:37
bechamelX0d_of_N0d: and should'nt be referenced by other modules, are they up to date ?21:37
X0d_of_N0dhold up21:37
X0d_of_N0dhg head21:38
X0d_of_N0dchangeset:   1332:0a5d9fa9c3e121:38
X0d_of_N0dbut I'm running it against a db that has an old version in it too21:38
cedkX0d_of_N0d: each modules21:38
X0d_of_N0dI just cloned all the modules in
X0d_of_N0dhey, is pastebin down?21:47
X0d_of_N0dguess not anymore...21:49
X0d_of_N0dhere are the versions for all my moduels
X0d_of_N0das far as I can tell they're all up to date21:55
bechamelX0d_of_N0d: try "grep relationship */" in the modules directory this could give you the culprit21:57
X0d_of_N0dhum... for some reason account didn't get installed but account_be did. I removed account_be and it seems to be working21:58
X0d_of_N0dlemme double check21:58
X0d_of_N0dI guess it's just that I didn't have account checked out...22:02
bechamelX0d_of_N0d: yes it seems that account is missing on your list22:05
X0d_of_N0dyeah, I thought it was installed but I guess I missed that22:07
X0d_of_N0dIt would be nice to set up the modules area as a forest so you could just check out everything all at once instead of having to get all the modules yourself22:08
bechamelX0d_of_N0d: there is a script that can do this work but i don't remember the url22:10
bechamelX0d_of_N0d: ced talk about it on the mailing list maybe it would be a good idea to put a link on the wiki too22:11
X0d_of_N0dyeah, or on the downloads page22:12
bechamelX0d_of_N0d: even better22:12
X0d_of_N0dI do kind of like the module repository idea tinyerp has, although the actual reposity is an absolute disaster...22:13
X0d_of_N0dbechamel: are you guys trying to avoid that kind of thing?22:16
bechamelX0d_of_N0d: actually something like intuxication or freehg or any online hg host should do the truck22:16
bechamelX0d_of_N0d: yes we try to avoid absolute disasters :D22:16
X0d_of_N0dIt seems like the major mistake tinyerp made was dropping everything into their repository22:17
X0d_of_N0ddo you disagree?22:17
bechamelX0d_of_N0d: you talk about launchpad or about the modules on the website ?22:18
bechamelX0d_of_N0d: anyway the problem with both is that they are to confusing, even experimented devs are sometimes lost22:19
X0d_of_N0dI'm talking about a way to import modules directly through the tryton interface rather than using hg.22:19
bechamelX0d_of_N0d: oh ok22:20
X0d_of_N0dbechamel: oh, I agree... I was trying to use an ldap module and found out that it was broken and the sourcecode was essentially gone22:20
bechamelX0d_of_N0d: one solution is to use the hg api from the client :)22:20
X0d_of_N0dthen you could easily put in alternate repos and have hg do all the work22:20
X0d_of_N0dand furthermore you'd get the benefit of still being able to do the same thing from the commandline22:21
bechamelX0d_of_N0d: or one can create a module "sync_modules_from_hg" :D22:21
X0d_of_N0dwhich would actually help me out since I'm having to do custom modules for the company I work for22:22
X0d_of_N0dand the workflow really sucks for tinyerp22:22
bechamelX0d_of_N0d: the problem with this kind of automation is even they are great for devs  they are not welcomed in production22:22
bechamelX0d_of_N0d: imagine this functionality on the demo server22:23
X0d_of_N0dbechamel: syncing moduels from the latest tip would be a problem, syncing with a known-good tag might work though22:23
X0d_of_N0dit could be up to the admin to determine how to use it22:24
bechamelX0d_of_N0d: for me has a developper i prefer to use the script once and after clone from time to time new module22:25
bechamelX0d_of_N0d: the same way i update/install module on the command line22:26
X0d_of_N0dbechamel: this could be used as a distribution system whereby bugfixes could be made easily accessable to users22:28
X0d_of_N0dadmin could be informed "there is an update available for module x, to fix this or that"22:29
X0d_of_N0dthey could choose to update or not22:29
X0d_of_N0dand all of that would be built in to the hg api22:29
bechamelX0d_of_N0d: yes maybe. If this functionality were available you would prefer it over fpull or easy_install or the shell script ?22:30
X0d_of_N0dand if they wanted to use external repos, no problem... put in the url and let hg take care of the rest...22:30
X0d_of_N0dit doesn't have to be over, it could be along side22:30
bechamelX0d_of_N0d: yes it's an option, but if nobody use it ...22:31
X0d_of_N0dNot everyone is commandline oriented.22:32
X0d_of_N0djust like not everyone is gui oriented22:32
X0d_of_N0dI'd probably use both22:33
bechamelX0d_of_N0d: ;)22:34
X0d_of_N0dbut I know my boss would like a gui, and most of the people I work with would like a gui22:34
X0d_of_N0dAnyway, it's no big thing... really shouldn't be a lot of code, so if I want it I could probably code it22:35
X0d_of_N0dI've been wanting to hack up hg for a while anyway22:35
X0d_of_N0dACTION is back22:43
-!- juanfer(n=juanfer@ has joined #tryton22:43
nicoe might be of interest for you23:21
bechamelnicoe: yes thanks we saw it this afternoon, but we didn't fell it worth an answer23:24
nicoeYes I though exactly the same way but it might be interesting to see how fp replies23:24
bechamelnicoe: yes i don't know23:25
nicoedon't know what ?23:25
cedknicoe: there is some arguments that are right and some others wrong23:25
bechamelnicoe: maybe you can suggest to fp to use relario, it works well :)23:25
nicoeI think he really doesn't want to try it23:27
bechamelnicoe: i don't know if it will be interesting, i don't like the idea to talk badly about openepr on his own forum (even if all i would say is correct)23:27
nicoebut I do not know if using an headless really slows down the process23:28
cedkwe can perhaps correct some errors about the interpretation of the list arguments23:28
bechamelof course  fp doesn't want to try it but it should be interesting :)23:28
nicoeLike bechamel I don't think there is any gain to reply to him23:28
nicoemoreover because this is on his forum23:28
bechamelabout slowness: nothing force you to use pdf you can open odt files directly23:29
bechamelbut it should be intersting to test very large reports23:30
nicoeI made some tests a while ago and relatorio was quicker but I did not use OO it was just the file generation that was tested23:30
nicoeI'll try to make some benchmarks in the next of the week23:30
nicoeI may have missed something but I don't think that openerp supports odt (only sxw afaik) ... as you 're monitoring their change more closely did you see something ?23:31
cedknicoe: they still use sxw -> rml -> pdf23:32
nicoeok thanks.23:32
cedkby the way, I try a report with ~1000 pages, the generation was quite faster (maybe 5-6 secs)23:33
cedkthe slowest was Ooo, that took ~20 sec to open23:33
bechamelcedk: with tables ? reportlab doesn't work well with big tables23:33
cedkbut you will have the same with any pdf file with the same amount of page23:34
cedkbechamel: it is with relatorio23:34
cedkI know that reportlab is very much slowest23:34
bechamel yes i know but as we know that it's very difficult to create big report with long tables on reportlab it would be interesting to test the same with relatorio+oo23:35
nicoereportlab is doing a lot of stuff that are not necessary so I suspect their method is the slowest. But the main bottleneck is probably the network here23:36
cedknicoe: no, it was not a so big file23:36
cedknicoe: the bottleneck is Ooo that need to allocate all the space for the pages23:37
nicoeThere are a lot of thing to test: big tables / big tables with computation inside them / big tables with conditions / etc.23:37
nicoebut the fact that their reporting solution is not easily unpleaggeable from tiny makes it difficult to test ...23:38
nicoecedk: ok23:38
cedknicoe: any way, generating the file will always be faster with relatorio than with reportlabe, for the simple reason that relatorio doesn't do any formatting23:38
cedkand by the way, we are not linked to Ooo as odt is a standard format, there is many software that read odt file23:40
nicoeof course23:40
nicoebut you are using oo to render it as a pdf, no ?23:41
cedkit is just a trick23:41
cedkfor me the pdf is not needed as odt is ISO23:41
nicoeyeah but fabien is using it to say that this is slower23:42
cedkand pdf not23:42
cedknicoe: yes, but what do you want to do23:42
cedknicoe: it is like the limit of 2000023:42
cedknicoe: he say you can increase the 80 to 2000023:42
cedkbut it will put down teh OpenERP client23:43
cedkor like saying that if you restart the Tryton server, the client crash23:44
nicoeyes I know but in the eye of the world he has a good answer :(23:44
cedkit is false but how do you prove it23:44
cedkany way if we answer he will remove surely the comment as he did already on previous post on news23:45
nicoeI can always make a benchmark on relatorio/reportlab/openreport and post it on my blog ...23:46
cedknicoe: yes you can :-)23:47
cedknicoe: but I think that people who will read your post, already think that relatorio is better23:47
nicoeOn planet python there is still a lot of people that do not know anything about other solution than reportlab and so on ...23:48
cedknicoe: ha yes, you blog is no planet python23:49
nicoeyep :) I'll try to do that ... That's a nice topic for a blogpost (with a lot of trolling possibilities :)) and I haven't post for weeks23:50
cedknicoe: maybe you can try to convert it into pdf with the openoffice.interact23:51
cedknicoe: I'm pretty sure that with a big table, reportlab will fail23:52
CIA-54tryton: matb roundup * #650/Items reappearing untranslated in new database: works for most items, but not for: "Products" "" "False" "product.product" "Deutsch" "Sicht" "0" "stock"23:52
cedknicoe: because reportlab try to compute the height of each row and it takes a lot of time23:52
cedknicoe: but there is a patch in OpenERP, to avoid that and it takes the first row as the base of the height23:53
nicoeI made myself a todo note to blog about that ...23:55
nicoeThe nice thing with opendocument is that by default it is openoffice that does the rendering ...23:58

Generated by 2.11.0 by Marius Gedminas - find it at!