IRC logs of #tryton for Tuesday, 2009-02-24 #tryton log beginning Tue Feb 24 00:00:01 CET 2009
-!- yangoon1( has joined #tryton00:03
-!- ChanServ(ChanServ@services.) has joined #tryton01:28
-!- juanfer(n=juanfer@ has joined #tryton02:13
-!- ikks(i=igor@ has joined #tryton02:42
-!- gremly(n=oscar@ has joined #tryton03:07
-!- gremly(n=oscar@ has joined #tryton04:05
-!- yangoon( has joined #tryton05:19
-!- paola( has joined #tryton07:46
-!- sharkcz( has joined #tryton07:47
CIA-10tryton: Timitos roundup * #810/problems with extending selection fields: [new] On this place in the documentation you describe how ...08:18
-!- johbo( has joined #tryton08:41
-!- nicoe( has joined #tryton09:08
-!- simahawk( has joined #tryton09:17
-!- simahawk( has left #tryton09:17
-!- Gedd( has joined #tryton09:39
-!- cristi_an(i=5978d3ce@gateway/web/ajax/ has joined #tryton09:41
-!- cedk(n=ced@gentoo/developer/cedk) has joined #tryton09:42
-!- kleinerdrache( has joined #tryton10:03
-!- tekknokrat( has joined #tryton10:09
-!- tekknokrat( has left #tryton10:10
-!- tekknokrat( has joined #tryton10:27
-!- bechamel(n=user@ has joined #tryton10:37
cristi_ancedk: thx for your answer10:42
cristi_anso making tha possible what is needed to be done10:44
cristi_ani can't tell to those customers do not do those reports simultaneously ?10:44
cristi_anor python is the fault10:44
cedkcristi_an: I don't know, you must use the right hardware for your needs10:46
cedkcristi_an: it will be always possible to overload the server with bad practice10:47
cristi_anwell here is an issue you can;t name bad practice 20 people that want some reports...10:49
cristi_anthey just use the program...10:49
cedkcristi_an: no, I want to say that you can at any time overload any server10:50
cristi_ani wonder for cases like this nothing can be done on server side10:50
cedkcristi_an: and as I said it is a matter of probability10:50
cristi_anwhen you do a query ...ansd iterate the result set adn pass data to reports10:51
udonocristi_an: buy Memory for $2010:51
yangoonudono: :D10:51
cristi_anto make somehow data that is already pased to the reports to be garbage colected or so10:51
cristi_ani do not knwo how relatorio works10:52
cristi_anjust asking10:52
cedkcristi_an: any way, you must have at least enough memory to have reports result in memory10:54
udonocristi_an: yes, cedk could implement a fifo dispatcher for XMLRPC requests, but I think not for $2010:54
cristi_anudono: you are so funny :)10:54
cristi_anudono: you mean like one request at the time10:55
cristi_anthat is messagin10:55
cristi_anlike a queue10:55
cristi_anbut that can be done for reports ....10:55
cristi_annot for regular inserts updates etc10:56
cristi_anwhy not10:56
udonocristi_an: relatorio does this afaik: Open the odt document, parse the content.xml files with genshi templating system, zip the files back to the newly parsed odt. nicoe is that right in general?10:56
nicoeudono: yes it is10:56
udononicoe: thanks, btw: hi :-)10:56
cristi_anudono: when data is filled ?10:56
nicoeudono, hello10:57
udonocristi_an: I don't see any pros on this only cons10:57
cristi_anudono: but that is not a bad ideea10:57
cristi_anrequest for reports to be queued10:57
cristi_annot run in parallel10:58
cristi_anfor very large reports...10:59
cristi_anbut that is harder then buying some memory :))10:59
udonocristi_an: it is a bad Idea, I think it makes everything complicated. dispatching and scheduling is the work of the OS, so better buy 8G Memory. Its more easy and scalable10:59
cedkcristi_an: why? we have a fast report generator10:59
cristi_ani have to admit you are right ...i am still in the window world where i jave limit in using RAM's11:00
cristi_anbut assume you will once have giant coporate running tryton11:01
cristi_anyou will need some load balaincing11:02
udonoACTION cited Laotse: When we have the one side, then we will have the other side, too.11:05
cristi_anyou are not that far ...on cebit you'll make ... waves :)11:07
udonocristi_an: we will see11:07
cristi_anudono: thx for open my eyes with that simple solution (for linux) buy ram...11:09
cristi_anthat will solve my problem for sure :)11:09
cristi_anbut i still wonder in the 95-2000 period how they handled this kind of problems ...11:17
cristi_ansince back then i do not htink there was like no limits on ram...stuff11:18
-!- oversize( has joined #tryton11:19
udonocristi_an: I see no benefit to implement every good-idea-which-could-be-usefull-in-future (TM). But I think if we have the real case someone really have an issue with scalability of tryton, it will be solved. But for now its something like a premature optimisation. Some one who will serve 100 Clients with a possible hard load on reporting knows that an erp-Implementation will cost minimum $100.000.11:28
cristi_angood-idea-which-could-be-usefull-in-future (TM).  LOL :)11:31
cristi_anfor me is ok ...the ram solution11:36
udonocedk: which module I need to install to use spellchecker on client side?11:41
udonocedk is not there...11:42
bechameludono: it's a gnome module iirc11:42
udonogtkspell I fond11:43
bechameludono: yes11:44
udonopython-gnome2-extras is the package...11:47
udonoon a kde system it needs 125MB extra. This seems a big dictionary for spell checking :-)11:47
bechameludono: yes, happilly it's not mandatory to run the client11:49
-!- ChanServ(ChanServ@services.) has joined #tryton11:54
-!- ChanServ(ChanServ@services.) has joined #tryton12:05
Timitosbechamel: i am working on a module thats has a one2many-field and in this field there is another one2many field. i have an issue when creating records with this constellation:
Timitosbechamel: the funny thing. i think yesterday it worked12:16
Timitosbechamel: could there be an issue with some changes yesterday?12:16
CIA-10tryton: udo.spallek * r447 /wiki/ Edited wiki page through web user interface.12:18
bechamelTimitos: there was changes for m2o12:18
bechamelTimitos: there is a function field in your model ?12:19
Timitosbechamel: yes12:19
bechamelTimitos: now a function field cannot return [(1,"val1"),(2,, "val2")] but must return {1:"val1", 2:"val2"}12:21
Timitosbechamel: it seems that when i try to save the record the required one2many relation field is not set properly12:21
Timitosbechamel: i get something like NULL-VALUE in column x violates not-null-constraint ....12:22
Timitosbut this shouldn´t happen when i am working on a one2many view?12:22
bechamelTimitos: a required one2many  ?12:22
Timitosbechamel: yes. but only for some states12:23
Timitosabout the function field. i will check this.12:23
bechamelTimitos: try to save the form with only data in the "first" on2many and then after add data to the "second" one2many (one2many are saved when the parent form is saved)12:24
Timitosbechamel: i tried. same problem.12:25
bechamelTimitos: btw, the error you pasted is about function field12:26
Timitosah. ok. so i think i have two different issues12:27
bechamelTimitos: yes maybe12:27
bechamelTimitos: the error appear when you save your form ?12:28
Timitosbechamel: yes12:29
Timitosbechamel: now i remember. yes i habe another issue with this module12:29
Timitosbechamel: i have a function field that depends on fields in the one2many field. if i add a new line in the one2many the function field is not updated in the upper view.12:30
Timitosbechamel: i think this is connected with the error message12:30
bechamelTimitos: so maybe you have only one problem because saving a form trigger a read just after (to refresh function field notably)12:31
bechamelTimitos: if you want to update functin field before saving the form you can use an on_change_* function12:32
Timitosbechamel: mom. i will try something12:34
bechamelTimitos: on_change_lines on account_invoice/ do a similar work12:35
Timitosbechamel: i will take a look thx12:35
Timitosbechamel: i do not have the function field that needs to be computed in the form view but only in the tree view. is this perhaps the problem?12:49
bechamelTimitos: for the on_change maybe, if not it should be a problem (function field are computed on the server, which is view-agnostic)12:52
bechamelshould/shouldn't/ :-/12:53
bechamelTimitos: the first thing to do is to print the result your computed field (check that there is 1 as key in the dict)12:58
Timitosbechamel: i think i found the problem. just a minute i am testing12:58
Timitosbechamel: found it. my function for the function fields didn´t return values for every id. thx for being with me13:03
bechamelTimitos: great13:04
-!- simahawk( has joined #tryton13:49
-!- simahawk( has left #tryton13:49
-!- kleinerdrache_( has joined #tryton14:07
-!- panthera( has joined #tryton14:40
-!- tekknokrat( has left #tryton15:20
-!- ikks(n=igor@ has joined #tryton15:31
-!- nicoe( has joined #tryton15:38
CIA-10tryton: matb roundup * #811/Modules: Refresh list automatically: [new] After installation of modules the list view should be refreshed automatically to reflect the real installation state.15:39
-!- juanfer(n=juanfer@ has joined #tryton15:59
CIA-10tryton: Bertrand Chenal <> default * 132:3b75f904462b sale/ ( sale.xml): Improved invoice exception handling16:40
CIA-10tryton: Bertrand Chenal <> default * 133:e44216953830 sale/ Typo16:40
CIA-10tryton: Bertrand Chenal <> default * 134:cccfe735ed2b sale/ ( sale.xml): Improved packing exception handling16:40
-!- cedk(n=ced@gentoo/developer/cedk) has joined #tryton17:16
cedkbechamel: how do you want to name _rpc_allowed?17:23
bechamelcedk: the problem is to mix two concept in one dict17:24
bechamelcedk: the same dict tell if rpc is allowed and if the cursor should be commited17:25
bechamelcedk: so there is not good name17:25
cedkbechamel: it must be in the same dict otherwize we will forget to set value for somes17:26
bechamelcedk: I think about {'read': {'commit': True}, ...} but it's a bit overkill17:29
cedkbechamel: it could be [('read', False), ('write', True)]17:30
cedkbechamel: it is will be less efficient when searching for one method value17:30
bechamel[('read', False), ('write', True)] is not better17:32
bechamelrpc_allowed = {'read': {'commit': True}, ...} is self-documented: one can call read and one should commit after that17:33
bechamelcedk: but if you prefer to keep it like that it's ok, it was just a suggestion17:34
cedkbechamel: but it doesn't force you to put commit values17:34
bechamelcedk: when I see rpc_allowed = {'read': False, 'write': True}, I read "rpc is allowed on write but not on read"17:40
cedkbechamel: why not _rpc_allowed_with_commit17:41
bechamel_rpc_allowed_but_the_value_for_each_key_is_to_tell_if_there_shoud_be_a_commit :D17:45
bechameljoke aside ..17:45
bechamelwhat if i want to tell that a method should do a commit but i don't want to allow rpc on it ?17:46
bechamelcedk: like a method call by a cron  ?17:46
cedkbechamel: you don't need to tell this17:47
bechamelcedk: and if a method should _not_ do a commit but i don't want rpc on it ?17:49
cedkbechamel: method will never do commit17:56
bechamelcedk: why ? what about the cron that create the purchase requests ?17:58
cedkbechamel: it is the cron that open the cursor and commit it, never any method17:59
-!- ChanServ(ChanServ@services.) has joined #tryton18:09
-!- vengfulsquirrel( has joined #tryton18:52
-!- bechamel(n=user@ has joined #tryton19:29
-!- cedk(n=ced@gentoo/developer/cedk) has joined #tryton19:40
-!- enlightx( has joined #tryton20:16
-!- udono( has joined #tryton20:31
-!- bechamel`(n=user@ has joined #tryton20:51
-!- vengfulsquirrel( has joined #tryton21:11
-!- paola_( has joined #tryton22:10
-!- Timitos( has joined #tryton22:19
-!- Timitos( has left #tryton22:20
CIA-10tryton: C?dric Krier <> default * 1193:cc9969de15b2 tryton/tryton/gui/window/view_form/view/form_gtk/ Improve check for float encoding with digits for issue80622:50
CIA-10tryton: ced roundup * #806/Calculation of amount in purchase/sale lines: [resolved] Fix with changeset cc9969de15b222:50
CIA-10tryton: C?dric Krier <> default * 1194:d6e1d1d42c3a tryton/tryton/ (3 files in 2 dirs): Add check on integer and float in list view22:51
CIA-10tryton: C?dric Krier <> default * 1195:5ba7613c231c tryton/tryton/gui/window/ Fix title_results in win_search when form name contains %23:01

Generated by 2.11.0 by Marius Gedminas - find it at!