IRC logs of #tryton for Friday, 2010-11-12 #tryton log beginning Fri Nov 12 00:00:01 CET 2010
2010-11-12 00:35 -!- mhi1( has joined #tryton
2010-11-12 01:26 -!- digitalsatori(~tony@ has joined #tryton
2010-11-12 01:43 -!- sharoon(~sharoon@ has joined #tryton
2010-11-12 03:50 -!- juanfer(~juanfer@ has joined #tryton
2010-11-12 05:18 -!- yangoon( has joined #tryton
2010-11-12 05:30 -!- sharoon(~sharoon@ has joined #tryton
2010-11-12 05:45 -!- digitalsatori(~tony@ has joined #tryton
2010-11-12 06:17 -!- digitalsatori(~tony@ has joined #tryton
2010-11-12 06:54 -!- zodman(~zodman@foresight/developer/zodman) has joined #tryton
2010-11-12 10:50 -!- irclog( has joined #tryton
2010-11-12 10:52 <cedk> is back
2010-11-12 10:53 <paepke> cedk, thanx
2010-11-12 10:55 <cedk> still memory issue
2010-11-12 10:56 <paepke> cedk, damaged or too less recources?
2010-11-12 10:57 <cedk> paepke: not enough memory
2010-11-12 10:58 <cedk> Timitos, udono, yangoon, paepke: did you plan to publish news about the release in the German "press"?
2010-11-12 11:11 -!- heg( has joined #tryton
2010-11-12 12:18 -!- woakas(~woakas@ has joined #tryton
2010-11-12 12:22 -!- lem0na(~lem0na@ has joined #tryton
2010-11-12 12:37 <yangoon> cedk: question: could we have another mailing list just for announces?
2010-11-12 12:38 <yangoon> because it is kind of common use to announce by mail, not only news or rss
2010-11-12 12:39 <cedk> yangoon: I find that is mailing list are not done for announce
2010-11-12 12:40 <yangoon> cedk: they are often used for this purpose and it is for sure the most widespread and reliable medium
2010-11-12 12:41 <paepke> yangoon, +1
2010-11-12 12:42 <dba> +1
2010-11-12 12:42 <cedk> so it will be just rss2email
2010-11-12 12:43 <paepke> cedk, why not. but it should be run at youre server to have the right from. and it would be anyway easier to subscribe to a tryton-announce mailinglist and push the rss2mail script to that mailing list
2010-11-12 12:43 <yangoon> I understand correctly: you would just set up an rss2email to a read-only list?
2010-11-12 12:44 <yangoon> this would fit perfectly
2010-11-12 12:44 <paepke> cedk, subscribing to mailing list is common. ( or google group maybe )
2010-11-12 12:45 <cedk> yangoon: yes
2010-11-12 12:45 <yangoon> cedk: great
2010-11-12 12:45 <dba> cedk: thanks
2010-11-12 12:48 <paepke> cedk, thanx
2010-11-12 12:50 <cedk> it will bring more memory consomation on the server :-)
2010-11-12 12:51 <yangoon> cedk: :)
2010-11-12 12:52 <paepke> cedk, maybe outsourcing the tryton-demo servers? i have some space left on one of my root servers.
2010-11-12 12:53 <cedk> paepke: demo is already on separated server
2010-11-12 12:53 <cedk> paepke: with unittest
2010-11-12 12:53 <cedk> I plan to get a new host next year
2010-11-12 12:55 <paepke> cedk, ok.
2010-11-12 13:04 -!- tony_(~tony@ has joined #tryton
2010-11-12 13:44 -!- dba(~daniel@ has joined #tryton
2010-11-12 13:52 <paepke> cedk, about that proteus issue: its kinda strange. against a fresh 1.8 it works. i maybe messed something up. sorry bothering you with the issue.
2010-11-12 13:53 <cedk> paepke: ok
2010-11-12 14:08 -!- hoRn_( has joined #tryton
2010-11-12 14:08 <hoRn_> hi
2010-11-12 14:12 <hoRn_> a little db question (i'm not familar with postgres): which action can trigger a ROLLBACK. in a modul what is connected to a webservice i have to update records each time the tryton-view is called: i check someting and update the lines. Logging postgres shos me a ROLLBACK immediately after the transaction
2010-11-12 14:13 <cedk> hoRn_: Tryton rollback every request except if it is defined on _rpc
2010-11-12 14:13 <cedk> hoRn_:
2010-11-12 14:15 <hoRn_> ok - so i have to set in in _rpc.update - cool
2010-11-12 14:17 <cedk> hoRn_: for each Model you want to track
2010-11-12 14:18 <hoRn_> cedk: ok - thank you
2010-11-12 14:18 <cedk> hoRn_: but be careful that it will decrease the performence of Tryton
2010-11-12 14:18 <hoRn_> cedk: ic can do it explicit in __init__, right?
2010-11-12 14:19 <cedk> hoRn_: yes
2010-11-12 14:20 <hoRn_> i did it with buttons before, not knowing exactly what i'm doing ;)
2010-11-12 14:20 <hoRn_> know i get it
2010-11-12 14:21 <hoRn_> now
2010-11-12 14:29 <paepke> hoRn_, cedk. following your discussion I don't get it what the rollbacking or not rollbacking with that _rpc is used for?
2010-11-12 14:31 <cedk> paepke: if the value in _rpc for a method is set to True then the transaction will be commited otherwise it is rollbacked
2010-11-12 14:31 <paepke> cedk, but normally i want my transaction to be commited?
2010-11-12 14:32 <paepke> cedk, its just to make a method available via rpc?
2010-11-12 14:34 <cedk> paepke: many call doesn't require to be commited
2010-11-12 14:34 <cedk> paepke: this behavior allow to prevent bad programing practice
2010-11-12 14:35 <paepke> cedk, ic
2010-11-12 14:36 -!- tony_(~tony@ has joined #tryton
2010-11-12 14:42 <hoRn_> cedk: my construction is like this: a function-field calls a method, this method calls the create and delete of another model - so what i need to set in the _rpc to True?
2010-11-12 14:44 <cedk> hoRn_: what does the function-field?
2010-11-12 14:46 <hoRn_> cedk: fields.Function(fields.One2Many()) - it calls a webservice and updates or delete records in the related model
2010-11-12 14:46 <cedk> hoRn_: fields doesn't call webservice
2010-11-12 14:46 <hoRn_> the method
2010-11-12 14:47 <cedk> hoRn_: I guess you mean when reading the Function field
2010-11-12 14:47 <hoRn_> cedk: yes
2010-11-12 14:48 -!- pheller(~pheller@ has joined #tryton
2010-11-12 14:50 <hoRn_> fields.Function(fields.One2Many('coating.configuration.line', 'configuration', 'saved hourly rates'),'get_some_informations_from_a_webservice_and_manipulate_coating_configuration_line'): return {rec_id:[ids_of_the_related_models]}
2010-11-12 14:51 <cedk> hoRn_: I think it is not a correct design
2010-11-12 14:51 <cedk> hoRn_: why do you want to modify records when reading?
2010-11-12 14:53 <hoRn_> cedk: in a external webservice my customer can create and manipulate settings, which i need for further operations in tryton,. so i read the actual settings from the external service and want to persist them in tryton
2010-11-12 14:56 -!- tony_(~tony@ has joined #tryton
2010-11-12 14:56 <bechamel> hoRn_: what about a cron job ?
2010-11-12 14:57 <cedk> hoRn_: yes cron or trigger should be better
2010-11-12 14:59 <hoRn_> cedk: its an idea - but a little bit unconnected to the real workflow. the client finally calls the external webservice with the id of the setting, then a external website is opening. There the cleint can manipulate the settings to his wishes and i need the information about this setttings
2010-11-12 15:01 <cedk> hoRn_: so it is a wizard flow
2010-11-12 15:01 <hoRn_> i will take a look to the trigger related stuff - the design above was my first idea - but the ROLLBACK delete the record
2010-11-12 15:03 <hoRn_> cedk: i think i have to rethink the design
2010-11-12 15:03 <cedk> hoRn_: I don't speak about the Tryton trigger
2010-11-12 15:03 <cedk> hoRn_: how the settings are entered?
2010-11-12 15:04 <hoRn_> cedk: at an external website
2010-11-12 15:05 <hoRn_> cedk: sometimes. we build the possibility to write records to the external webservice. but the webservice has only set and get method - there is no delete or change method
2010-11-12 15:06 <hoRn_> cedk: so we need to update our records, because the clent NEED to manipulate them outside of tryton
2010-11-12 15:10 <cedk> hoRn_: still don't understand
2010-11-12 15:11 <cedk> hoRn_: you said that the client needs to enter setting to call the webservice and because of that you can not make a cron
2010-11-12 15:12 <hoRn_> the settings to call the webservice are stored in a singleton.
2010-11-12 15:13 <cedk> hoRn_: so you can run a cron job to sync data
2010-11-12 15:13 <hoRn_> cedk: yes - i will take a look - its sound better
2010-11-12 15:14 <hoRn_> cedk: i was trying to update them on the fly - not than important
2010-11-12 15:14 <cedk> hoRn_: because otherwise you need to set read method of parent model commited
2010-11-12 15:15 <hoRn_> cedk: how ;)
2010-11-12 15:15 <cedk> hoRn_: with _rpc
2010-11-12 15:16 <cedk> hoRn_: but you should do that for every entry point method that could call the function field
2010-11-12 15:16 <hoRn_> cedk: rpc.update('read',True)
2010-11-12 15:17 <cedk> hoRn_: there is an other way, which is that you could create a new transaction inside the function field that will be commited
2010-11-12 15:18 <hoRn_> cedk: puh
2010-11-12 15:18 <cedk> hoRn_: the choice depend of the speed of retreiving data from the webservice
2010-11-12 15:18 <cedk> hoRn_: but you could be a kind of timeout on stored data
2010-11-12 15:19 <hoRn_> cedk:ok
2010-11-12 15:19 <cedk> I created
2010-11-12 15:20 <cedk> I'm just waiting for gmane approuval to store the history before running r2e
2010-11-12 15:21 <cedk> hoRn_: but if you can easily detect if data on the webservice has changed then a cron job could be the easiest way
2010-11-12 15:21 -!- pepeu(~manuel@ has joined #tryton
2010-11-12 15:22 <cedk> did someone know if it is possible to call a method on a descriptor?
2010-11-12 15:22 <hoRn_> cedk: yes - i think finally this could be nicer
2010-11-12 15:30 <bechamel> cedk: descriptor ?
2010-11-12 15:32 <cedk> bechamel:
2010-11-12 15:41 -!- cedk(~ced@gentoo/developer/cedk) has joined #tryton
2010-11-12 15:41 -!- tony_(~tony@ has joined #tryton
2010-11-12 15:46 <cedk> bechamel: I think I can use __dict__ of the class
2010-11-12 15:48 <bechamel> cedk: you want something like obj.__get__.my_method() ?
2010-11-12 15:49 <cedk> bechamel: no
2010-11-12 15:49 <cedk> Class.__dict__['field'].method()
2010-11-12 15:49 <cedk> bechamel: it is for proteus
2010-11-12 15:50 <cedk> I have added descriptor and now I would like to retrieve value in the format for update or on_change
2010-11-12 15:52 <bechamel> ok, and it works with the __dict__ ?
2010-11-12 15:53 -!- vladimir_(~vladimir@ has joined #tryton
2010-11-12 15:53 <cedk> bechamel: not yet tested
2010-11-12 16:58 -!- pjstevns( has left #tryton
2010-11-12 17:25 -!- vladimir_(~vladimir@ has joined #tryton
2010-11-12 17:34 -!- tony_(~tony@ has joined #tryton
2010-11-12 18:13 -!- zodman(~andres-va@ has joined #tryton
2010-11-12 18:50 -!- trifon( has joined #tryton
2010-11-12 19:03 -!- lem0na(~lem0na@ has joined #tryton
2010-11-12 19:38 -!- enlightx(~enlightx@ has joined #tryton
2010-11-12 19:45 -!- chrue( has joined #tryton
2010-11-12 19:59 -!- plantian( has joined #tryton
2010-11-12 20:26 -!- gavinf( has joined #tryton
2010-11-12 20:46 -!- pepeu(~manuel@ has joined #tryton
2010-11-12 20:47 -!- lem0na(~lem0na@ has joined #tryton
2010-11-12 21:10 -!- gordon_geeko( has joined #tryton
2010-11-12 21:14 <paepke> cedk, neso-1.8.0.tar.gz has only 61.2K ?
2010-11-12 21:14 <paepke> cedk, or should i talk to yangoon?
2010-11-12 21:15 <yangoon> paepke: what's wrong about it?
2010-11-12 21:15 <paepke> yangoon, its too small?
2010-11-12 21:16 <cedk> paepke: it is almost like others
2010-11-12 21:16 <paepke> ah youre right. but do i have to install all the modules by hand?
2010-11-12 21:17 <yangoon> paepke: yes, it is neso, not an all-in-one
2010-11-12 21:17 <paepke> i only used the windows version to test, where all modules are in that installer.
2010-11-12 21:17 <paepke> yangoon, cedk sorry for the misunderstanding. i were just wondering.
2010-11-12 21:18 <paepke> shouldn't be a neso all in one on linux, too?
2010-11-12 21:20 <paepke> i know that there is a difference between neso and the tryton/trytond combination ;-)
2010-11-12 21:22 <gordon_geeko> how do i replace an entire view when using an existing module?
2010-11-12 21:23 <gordon_geeko> i tried same name, higher priority and replace for xpath //form - both don't appear to be working
2010-11-12 21:24 <udono> gordon_geeko: no, we replace only parts when there is something to replace.
2010-11-12 21:25 <udono> gordon_geeko: how is the id of your view? and which view you want to replace?
2010-11-12 21:25 <gordon_geeko> udono: what if i want to change the entire form (account.move_view_form)
2010-11-12 21:26 <udono> gordon_geeko: then you are going to break modularity...
2010-11-12 21:27 <gordon_geeko> udono: i'm reusing the fields, but adding some to - and the layout is different
2010-11-12 21:27 <udono> gordon_geeko: maybe just create an alternate view and make a menu entry for your view. After this set the menu entry of the old view active eval="False"
2010-11-12 21:28 <gordon_geeko> udono: i'll give it a try, thanks!
2010-11-12 21:28 <udono> gordon_geeko: the problem is, when you want to install another module which do view modifications too.
2010-11-12 21:30 <gordon_geeko> udono: yes, good point - but for lots of changes would you recommend multiple changes via xpath?
2010-11-12 21:32 <udono> gordon_geeko: If the original view is this bad, it could be an idea to redesign it in general. We generally respect the original views in our development and modify them all with xpath.
2010-11-12 21:33 <gordon_geeko> udono: not bad, just missing fields and because of that i need to optimize the layout
2010-11-12 21:33 <gordon_geeko> udono: i'm gonna try the xpath approach first
2010-11-12 21:35 <udono> gordon_geeko: for me it is the best way. And if I would need really an alternative view, I would redesign it under another id then the original view, like told above.
2010-11-12 21:36 <gordon_geeko> udono: it is indeed a trade-off
2010-11-12 21:37 -!- chrue( has joined #tryton
2010-11-12 22:45 -!- hoRn_( has joined #tryton
2010-11-12 22:46 <hoRn_> sleepless in Leipzig
2010-11-12 22:46 <hoRn_> cedk: cron is the better design ;)
2010-11-12 22:58 <cedk> hoRn_: great
2010-11-12 23:51 -!- gremly(~gremly@ has joined #tryton

Generated by 2.17.3 by Marius Gedminas - find it at!