chat.freenode.net #tryton log beginning Thu Sep 29 00:00:02 CEST 2011 | ||
2011-09-29 01:51 <plantian> Hi, what does the new search widget do? | ||
2011-09-29 01:57 <plantian> oh I guess autocomplete, I guess tdp was the last file i checked out of multiple patches | ||
2011-09-29 03:11 -!- ciupicri(~ciupicri@unaffiliated/ciupicri) has joined #tryton | ||
2011-09-29 14:34 -!- pjstevns(~pjstevns@a83-163-46-103.adsl.xs4all.nl) has left #tryton | ||
2011-09-29 17:54 <sisalp> hello, | ||
2011-09-29 17:55 <sisalp> making a sale order I don't read the product description in the sale order. Am I missing something ? | ||
2011-09-29 17:56 <cedk> sisalp: not sure to understand | ||
2011-09-29 17:57 <sisalp> I'm used to oe, where product descriptions are used in order lines notes | ||
2011-09-29 18:00 <cedk> sisalp: I test it here and I got the description filled when I select a product in a sale line | ||
2011-09-29 18:01 <sisalp> ok, so I 'll make another try | ||
2011-09-29 18:01 <sisalp> the description is filled with the product name | ||
2011-09-29 18:02 <sisalp> while notes are not filled with product description | ||
2011-09-29 18:02 <sisalp> this was the point | ||
2011-09-29 18:02 <cedk> sisalp: note is a free text | ||
2011-09-29 18:03 <cedk> sisalp: and the goal of note is not to show it to the customer | ||
2011-09-29 18:04 <sisalp> ok, so shouldn't decription to be filled with product decription if not empty ? | ||
2011-09-29 18:05 <sisalp> and product name if description is empty ? | ||
2011-09-29 18:06 <cedk> sisalp: I don't think | ||
2011-09-29 18:06 <cedk> sisalp: I don't want to have on a sale order or invoice the manual of the product | ||
2011-09-29 18:06 <sisalp> so how can I differentiate the name for the client and the supplier from the internal name ? | ||
2011-09-29 18:07 <sisalp> and what is product description for ? | ||
2011-09-29 18:08 <cedk> sisalp: supplier name is on supplier info | ||
2011-09-29 18:08 <cedk> sisalp: we don't manage an internal name | ||
2011-09-29 18:10 <sisalp> ok, is it possible to differentiate the description between two suppliers ? | ||
2011-09-29 18:11 <cedk> sisalp: there is a code and a name per supplier | ||
2011-09-29 18:15 <sisalp> ok, my product was not buyable, it is ok | ||
2011-09-29 18:15 <cedk> sisalp: we show only the interesting info :-) | ||
2011-09-29 18:15 <sisalp> i understand this differntiation is only for suppliers and cannot be used for clients ? | ||
2011-09-29 18:16 <sisalp> or if a product can be bought, the client can be added to the list ? | ||
2011-09-29 18:16 <cedk> sisalp: no only for supplier | ||
2011-09-29 18:17 <cedk> sisalp: you know someone who is using a product name different per customer? | ||
2011-09-29 18:26 <sisalp> yes | ||
2011-09-29 18:27 <sisalp> local subcontractors use the name of their clients (peugeot, renault, Alcatel, big ones) | ||
2011-09-29 18:28 <sisalp> but this can be added as a module for them | ||
2011-09-29 18:39 <sisalp> cedk: by the wayyou made a product search including partner's reference for one of them in the past (on oe) | ||
2011-09-29 18:40 <cedk> sisalp: yes but it is really custom | ||
2011-09-29 19:19 <hoRn> cedk: can you check my commit if its done right? | ||
2011-09-29 19:21 <cedk> hoRn: which one? | ||
2011-09-29 20:41 <sharoon> cedk: ping | ||
2011-09-29 21:04 <cedk> sharoon: pong | ||
2011-09-29 21:05 <sharoon> cedk: just wanted to say hi ;) | ||
2011-09-29 21:05 <cedk> sharoon: did you see that OE has implemented your sequence module | ||
2011-09-29 21:05 <cedk> sharoon: in the core of OE | ||
2011-09-29 21:06 <sharoon> cedk: did not see, i don't follow OE developments | ||
2011-09-29 21:06 <sharoon> cedk: i hope they have a copyright notice | ||
2011-09-29 21:06 <cedk> sharoon: but enough different to not share the copyright :-) | ||
2011-09-29 21:06 <sharoon> cedk: oh i thought so | ||
2011-09-29 21:06 <cedk> sharoon: need to look at it carefully, to be sure | ||
2011-09-29 21:07 <cedk> sharoon: is it well going your project? | ||
2011-09-29 21:07 <sharoon> cedk: they are copycats as always | ||
2011-09-29 21:07 <sharoon> cedk: all well | ||
2011-09-29 21:07 <sharoon> cedk: i wanted to clarify something regarding ir_translations | ||
2011-09-29 21:07 <sharoon> cedk: the constraint in ir_translations is a crazy little query which takes over 5 minutes to execute on our DB server (and its a top spec one) | ||
2011-09-29 21:09 <cedk> sharoon: check_unique_model ? | ||
2011-09-29 21:09 <sharoon> cedk: yes | ||
2011-09-29 21:09 <sharoon> cedk: i did not understand clearly what it does too | ||
2011-09-29 21:10 <cedk> sharoon: I find the messsage clear | ||
2011-09-29 21:10 <cedk> sharoon: but for sure it is not very performent | ||
2011-09-29 21:11 <sharoon> cedk: do you want an explain analyze on that ? | ||
2011-09-29 21:11 <cedk> sharoon: it could perhaps been implement in postgresql with a partial index | ||
2011-09-29 21:11 <cedk> sharoon: why not | ||
2011-09-29 21:11 <sharoon> cedk: can you explain what this constraint is supposed to do ? | ||
2011-09-29 21:11 <cedk> sharoon: by the way, did you get your passport for the unconference? | ||
2011-09-29 21:12 <sharoon> cedk: not yet | ||
2011-09-29 21:12 <sharoon> cedk: they will send me an appointment | ||
2011-09-29 21:12 <sharoon> cedk: waiting for that | ||
2011-09-29 21:13 <cedk> sharoon: it ensures that translations of type 'model' are unique | ||
2011-09-29 21:13 <sharoon> cedk: because modules (and hence models) are dynamically discovered ? | ||
2011-09-29 21:14 <cedk> sharoon: Fix duplicate translation for issue1717 | ||
2011-09-29 21:14 <sharoon> cedk: looking at ticket | ||
2011-09-29 21:17 <cedk> sharoon: it is a check added to no more have the issue | ||
2011-09-29 21:21 <cedk> sharoon: but for sure the check should be improved (at least for pg) | ||
2011-09-29 21:22 <sharoon> cedk: this is from my logs. i am running explain analyze 00000LOG: duration: 669591.375 ms statement: SELECT count(1) FROM "ir_translation" WHERE type = E'model' AND res_id != 0 GROUP BY name, res_id, lang, type, src HAVING count(1) > 1 | ||
2011-09-29 21:23 <cedk> sharoon: at least you can desactivate it in production | ||
2011-09-29 21:23 <cedk> sharoon: how many rows in ir_translation ? | ||
2011-09-29 21:24 <sharoon> cedk: 8 million | ||
2011-09-29 21:24 <sharoon> cedk: i already disabled it in production | ||
2011-09-29 21:25 <cedk> sharoon: you can create partial index http://www.postgresql.org/docs/8.2/static/sql-createindex.html | ||
2011-09-29 21:25 <cedk> sharoon: submit an issue to have this check only on non-pg DB | ||
2011-09-29 21:26 <sharoon> cedk: sure | ||
2011-09-29 21:26 <cedk> sharoon: if it is your only issue with Tryton, it is wonderful :-) | ||
2011-09-29 21:26 <sharoon> cedk: well, this is the second scalability issue we found (first was the sequence which we already fixed) | ||
2011-09-29 21:26 <sharoon> cedk: we have never had any other issue :) | ||
2011-09-29 21:32 <sharoon> cedk: http://paste.pocoo.org/show/484692/ | ||
2011-09-29 21:33 <cedk> sharoon: yep partial index is the solution | ||
2011-09-29 23:29 <cedk> sharoon: one scalability issue, I know is when loading large XML file like in GNUHealth | ||
2011-09-29 23:29 <sharoon> cedk: its also because all that is done in one large transaction | ||
2011-09-29 23:30 <cedk> sharoon: no | ||
2011-09-29 23:30 <sharoon> cedk: how big is the xml file in GNU health ? | ||
2011-09-29 23:30 <cedk> sharoon: it is because we store a BrowseRecord for each XML record in a module | ||
2011-09-29 23:31 <sharoon> cedk: Browse records suck a lot of resources, i have been running our production server with weak refs for over a month now and it seems to work fine and overall memory consumption has dropped | ||
2011-09-29 23:31 <cedk> sharoon: arround 5M | ||
2011-09-29 23:32 <cedk> sharoon: even with the LRU patch? | ||
2011-09-29 23:32 <sharoon> cedk: 5M records or 5M lines ? | ||
2011-09-29 23:32 <cedk> sharoon: 5Mb | ||
2011-09-29 23:32 <sharoon> cedk: yes | ||
2011-09-29 23:32 <sharoon> cedk: ok | ||
2011-09-29 23:33 <cedk> sharoon: what did you do with weak refs? | ||
2011-09-29 23:34 <sharoon> cedk: the browse records don't get deleted (GC'd) till the transaction is stopped. When you use a weak ref in browse record to store the cursor the browse record object s deleted | ||
2011-09-29 23:36 <cedk> sharoon: ho, good tip | ||
2011-09-29 23:36 <sharoon> cedk: internally for performance we use csv reports which directly write to NamedTemoraryFiles rather than memory and we saw an interesting pattern in a stock report. The report is quite fast but slows down by over 15 times when we add a single column - the cost price | ||
2011-09-29 23:36 <cedk> sharoon: please submit a patch | ||
2011-09-29 23:36 <sharoon> cedk: sure | ||
2011-09-29 23:37 <cedk> sharoon: don't understand the report stuff | ||
2011-09-29 23:38 <sharoon> cedk: there is a huge performance issue when property field cost price is accessed, i haven't figured where the issue is though | ||
2011-09-29 23:39 <cedk> sharoon: it is a Function field | ||
2011-09-29 23:40 <sharoon> cedk: http://hg.tryton.org/2.0/modules/product/file/06b45ca903de/product.py#l30 | ||
2011-09-29 23:41 <cedk> sharoon: http://hg.tryton.org/trytond/file/199aa7a3c423/trytond/ir/property.py#l27 | ||
2011-09-29 23:41 <cedk> sharoon: it is this method which is slow I guess | ||
2011-09-29 23:42 <cedk> sharoon: but https://elveos.org/en/features/840/description?title=improve-property-fields | ||
2011-09-29 23:42 <cedk> sharoon: should improve it | ||
2011-09-29 23:42 <cedk> sharoon: because I think the slow comes from the processing of value | ||
2011-09-29 23:43 <sharoon> cedk: ok |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!