IRC logs of #tryton for Saturday, 2011-11-19 #tryton log beginning Sat Nov 19 00:00:01 CET 2011
cedksphinx support translation by default:
nicoeI wonder how sphinx splits the token for translation00:11
nicoe"Every single element in the doctree will end up in a single message"00:11
nicoewhat does that mean ?00:11
cedknicoe: I guess it is all text not separated by an empty line00:12
cedkand I hope :-)00:12
nicoeI am afraid that a "**test**" would be also a singe element00:13
bechamelnicoe: I think the rst lexer is a bit smarter than this00:14
bechamelnicoe: I just checked some generated html and the <p> tags are set correctly00:15
cedkACTION trying sphinx translation00:15
nicoeI tried with rst2pseudoxml and it works as I expected00:17
nicoeIt splitted the emphasis part from the other element of the paragraph00:18
nicoeThat's a bit stupid (and according to the man I saw the doctree)00:18
cedkit looks like what I said00:19
cedkso you get just one block to translate par paragraph00:19
cedkyou don't get the "========"00:20
cedkbut you have the ":ref:`genindex`"00:20
cedkyou don't get the comments00:20
nicoeBut the phrase like "I am *just* a test" ?00:20
nicoeIs it one piece or three ?00:20
cedkone piece00:21
nicoeNice then00:21
cedkso that's means translator should be careful to respect the syntax00:21
nicoeIt kinda contradict the output of rst2pseudoxml but that's more clever00:21
bechamelto bad for the :ref:`genindex`00:21
nicoeYes but sometimes in translation you can not put the emphasis on the same words but only on the same concept00:22
cedkbut you don't get the bullets :-)00:22
nicoeAnd the syntax is simple so I guess that they will learn it quite quickly00:23
cedknicoe: yes, the only issue is that translator need to test before commiting00:23
cedknicoe: because they can break the syntax and then sphinx will not compile00:23
cedkso one thing less to do for the website :-)00:26
mdmyrgrasbauer: are you working in product variants ¿?00:43
marc05is anyone working in product variants ¿?00:45
grasbauerneeds the function field for concatenate the many2many ;)00:49
marc05so it is not working like a wizard to define variants00:50
grasbauerwe lost in discussions about PEP8 ;) - I'm more PUNK7100:50
grasbauermarc05: there is an button on product.template to generate all possible variants according attributes set on the templates00:52
marc05good, maybe i can try it later00:53
-!- grasbauer( has left #tryton01:11
zxq9Strange. I am getting the "Incompatible version of the server" message -- but only when connecting from a remote client. A local client connects fine. Same version on both machines, even when firewall turned off.10:53
udonozxq9: Which version?10:55
zxq92.1, a Mercurial clone from about 3 days before it started being called 2.210:56
zxq9I'm going to try a new download tar of 2.2 and see if the same thing happens.10:57
cedknew proposal for the main page:
zxq9Odd. Running it from a user-owned directory gives "Could not connect to the server". Running from a group-owned directory gives "Incompatible version of the server" to the same server.11:08
zxq9Do any ports on the client need to be opened?11:09
sisalpcedk: about elveos, what is the situation on VAT ? I 'm afraid you won't get the amount indicated11:17
zxq9I can at least get a response (an error message) if I activate and try to connect to xmlrpc. jsonrpc isn't responding to the client request at all.11:23
cedksisalp: good question, normally they will provide us an invoice with the VAT number of B2CK11:23
cedksisalp: oops, that's for the contributors11:24
cedksisalp: for us, I guess we will invoice it without VAT because it is a service in intracom11:24
udonozxq9: sorry, no idea11:26
zxq9udono: Ever seen this: No JSON object could be decoded (connecting to xmlrpc)11:26
udonozxq9: No11:28
zxq9Is there any documentation on the trytond.conf file that might shed some light on this or give me new directions to troubleshoot?11:29
bechamelzxq9: first thing to do is check if the port is open11:30
bechamelzxq9: the end of the server should contains: INFO:server:starting JSON-RPC protocol on localhost:800011:30
zxq9Done. Actually overdone. iptables is off (first thing I thought too)11:30
zxq9benchamel: Server says that, yes. And local connection is no problem.11:31
bechamelzxq9: test number 2:
zxq9exact same11:32
bechamelzxq9: then if you create a db from the client nothing append ?11:34
zxq9From a client on the local machine everything works fine. I cannot connect to begin database creation from a remote client.11:34
bechamelzxq9: on which ip the server is listening ?11:35
zxq9ping via hostname or IP works fine.11:35
zxq9SSH is fine, and the machines share over NFS also. So there is not a network problem here.11:35
cedkbechamel: new proposal for the main page:
zxq9The version of client software is identical.11:36
bechamelzxq9: and the nc test from remote host ?11:36
bechamelcedk: the "finance it" should be white11:37
bechamelcedk: and the hline cross the button11:38
bechamelcedk: appart from that is far better than the previous one11:38
cedkbechamel: fixed11:39
zxq9nc to port 8000 gives me nothing. No response at all from remote.11:40
zxq9nc to 8069 gives me the correct response from the XML-RPC port, but that isn't working for connections.11:40
zxq9For some reason I can't connect to JSON-RPC... :-S11:41
zxq9Is there anything that would permit local JSON-RPC and forbid remote JSON-RPC?11:41
bechamelzxq9: another way to troubleshoot is to launch "python2 -m SimpleHTTPServer 8000" on the server machine11:41
bechameland see if nc can connect11:41
zxq9Yes, this works fine.11:42
cedkACTION I hate HTML11:42
zxq9I hate XML also.11:42
zxq9And I really hate when (XML + HTML + AJAX = New dev platform!) in the eyes of 1st year CS students.11:43
zxq9Who become huge software marketing executives later...11:44
bechamelACTION smile11:44
bechamelzxq9: then try to put in the trytond config11:44
zxq9As hostname or jsonrpc setting in place of "localhost"?11:45
cedkbechamel: I don't know why this *fucking* is to high11:46
bechamelzxq9: jsonrpc11:46
zxq9benchamel: This worked.11:46
zxq9Why did this work?11:46
zxq9benchamel: And THANK YOU!11:46
bechamelzxq9: when trytond open the port he used those params11:46
bechamelzxq9: after that it's just the way socket works11:46
zxq9Why does using the actual hostname not work to open remote connection?11:47
bechamelzxq9: this is obvious one you realized that a same server can respond to several ip11:47
bechamelthrough different nic or not11:47
zxq9it is just something I didn't think of because I'm not using a cluster or KVM or anything.11:48
zxq9but I have never used like this before.11:48
udonozxq9: I guess because of your LANG setting in environment.11:48
udonozxq9: :-)11:48
zxq9hmmm... I need to look into this. Of course these systems are ja_JP-UTF.8, so... hehe11:49
udonoACTION never good to scroll up and answer old questions...11:49
bechamelzxq9: when i asked you on which ip the server is listening, I was talking about this config value, I should have been more precise11:49
zxq9You've helped me a lot walking me through your troubleshooting.11:50
bechamelcedk: add a "min-height: 300" to  .elveos-compact-feature in the css to see if it helps11:51
cedkbechamel: done11:52
bechamelcedk: so it's the right thing to do11:52
bechamelcedk: just fine tweak the value11:52
bechamelI have to go, bye11:53
cedkbechamel: done11:55
sisalpcedk: you asked two days ago about the 0 price when ordering a free service, then left before I anwered12:26
sisalpcedk: the answer is that I can put a non 0 price because I don't make any invoice later (Manual)12:27
sisalpcedk: the webshop is ok with 0 prices12:29
cedksisalp: I think it can then be managed without the Manual option, if sale doesn't create invoice line for line with price 012:30
yangooncedk: I think invoice lines with price 0 should be created12:31
sisalpsome times we need a 0 pro-format invoice, we still have the limitation12:31
yangoonyou could need/want them to inform the customer12:31
sisalpcedk: I remember this 0 value was already discussed. What was the conclusion ?12:32
yangoonsisalp: it should be possible12:32
albertcaThinking about full text search discussion we had on TUL...13:49
albertcaI think performance would be very bad if we try to fill in the field from the orm13:49
cedkTimitos: about issue2233, I'm thinking if it is not linked to translation13:50
albertcabecause we should read the record on every write operation13:50
cedkalbertca: any way, for me the idea of a rec_name fulltext search is not very good13:51
Timitoscedk: for the moment i do not have any idea about the reason. i need to debug it step by step.13:52
albertcamaybe we should keep it like Koo, in which fts is an "extra" feature13:52
cedkalbertca: there is 2 things:13:52
cedkalbertca: 1 - searching in a natural way13:53
cedkalbertca: 2 - searching for any kind of record13:53
albertcacedk: I agree13:53
cedkalbertca: I think 1 is the first to fix13:54
cedkalbertca: and this should be done with trigram index to speed ilike search13:55
albertcacedk: yes, the other one is an "extra" feature13:55
albertcacedk: we must keep in mind that trigram index will probably be slower13:55
cedkalbertca: slower than what?13:55
albertcacedk: so it should be used only when necessary: not for stock moves for example13:56
albertcacedk: than normal indexes or no index13:56
albertcacedk: slower to build, I mean :)13:56
cedkalbertca: ok, I think it is just for Char/Text fields that have select=True13:57
cedkalbertca: I think also about reference field where there is often search like 'model_name,%'13:57
albertcacedk: how do you want to solve that?13:58
cedkalbertca: just improve TableHandler to add trg index on such fields13:59
albertcacedk: I mean, the reference field, issue14:00
albertcacedk: I think I get the problem you mean14:00
cedkalbertca: I'm not sure, we need to check but I'm wondering if postgresql could not use btree index for this kind14:01
albertcacedk: It rings a bell to me, that latest PostgreSQL versions are capable of optimize 'start with' operations in like without the trigram index14:02
albertcacedk: after all, if the index is a sorted tree, it should be a relatively fast operation14:02
cedkalbertca: I think I had read something like that14:02
Timitoscedk: about issue 2233: it looks like a dirty cache. the LRUDict looks very weird after a few accounts has been created. but i am not yet on the bottom of the problem15:52
cedkTimitos: which Python version?15:55
Timitoscedk: 2.6.615:56
cedkTimitos: so it should be the backport of ordereddict in trytond/tools/ordereddict.py15:57
cedkTimitos: as I don't have the issue with Python2.715:57
cedkTimitos: I will make a try15:57
Timitoscedk: yes i think that it is related to the LRUDict, too. I will continue my research16:00
Timitoscedk: could be also something which is related to a problem with LRUDict an parent<->child concept16:05
cedkTimitos: run with the bacport OrderedDict and no issue16:23
Timitoscedk: i have things like this in the LRUDict: (810, {'create_uid': {}, 'type': {}, 'parent': {}})16:32
Timitosi think this should not happen? am i right?16:33
cedkTimitos: no16:34
cedkTimitos: but what are you printing?16:34
Timitoscedk: i am printing self._local_data of a BrowseRecord16:34
cedkTimitos: have you a language activate on your database?16:39
Timitoscedk: german and english. admin user is set to german16:40
cedkTimitos: ok, I really thing the setLang is breaking something16:44
Timitoscedk: can you reproduce it?16:45
Timitoscedk: could be this line:
cedkTimitos: I setup a DB right now16:47
cedkTimitos: normally this line set just an empty dict for a record16:48
Timitoscedk: self._local_data does not have the model dict layer but self._cache has it16:59
cedkTimitos: what do you mean?16:59
Timitoscedk: self._cache dict structure is cache[model][record] and self._local_data dict structure is local_data[record]17:01
Timitoscedk: i think this is the reason for the wrong values in self._local_data17:01
cedkTimitos: seems you are right17:01
cedkTimitos: ok can reproduce it now17:02
cedkTimitos: so it is clearly the language17:02
Timitoscedk: yes17:02
cedkTimitos: I think I have a patch17:04
Timitoscedk: that sound great :-)17:05
cedkTimitos: submitted17:15
TelesightFor those interested:17:57
Telesight A new subject is available:17:57
TelesightWorkshop result Production module.17:57
TelesightAnd another subject18:16
TelesightEmail integration. See:
TelesightFor those that are interested: Video OLAP integration is available. See the wiki.19:24
TelesightOk all the video's of friday TUL2011 are on the Wiki now, nothing more to add.21:38
cedkTelesight: do you plan to upload also on vimeo?21:39
Telesightcedk; That is possible, only a lot of work and the upload limit of Vimeo gives a delay.21:42
-!- Telesight( has left #tryton23:29

Generated by 2.11.0 by Marius Gedminas - find it at!