IRC logs of #tryton for Monday, 2010-10-18 #tryton log beginning Mon Oct 18 00:00:01 CEST 2010
-!- gremly(~gremly@ has joined #tryton00:17
-!- yangoon( has joined #tryton05:18
-!- Timitos(~kp@ has joined #tryton07:29
-!- enlightx( has joined #tryton07:31
-!- gremly(~gremly@ has joined #tryton08:02
-!- rednul_( has joined #tryton08:34
-!- pjstevns( has joined #tryton09:29
-!- pjstevns( has joined #tryton09:30
-!- udono( has joined #tryton10:25
-!- cedk(~ced@gentoo/developer/cedk) has joined #tryton10:31
-!- evernichon( has joined #tryton10:54
-!- bechamel( has joined #tryton10:56
-!- Red15(~red15@unaffiliated/red15) has joined #tryton10:56
-!- mr_amit(~amit@ has joined #tryton11:52
-!- paepke( has joined #tryton11:58
-!- mr_amit(~amit@ has joined #tryton14:02
-!- pheller( has joined #tryton15:18
-!- heffer_( has joined #tryton15:54
-!- heffer_(~felix@fedora/heffer) has joined #tryton15:54
-!- pheller(~pheller@2002:ad30:d8c3:0:217:f2ff:fe04:afe) has joined #tryton17:03
-!- pjstevns( has left #tryton17:05
-!- pheller_( has joined #tryton17:18
-!- vladimir_(~vladimir@ has joined #tryton17:21
-!- paepke( has joined #tryton17:24
paepkecan't trytond be killed anymore using ctrl-c when using it as console-program (not running as daemon)?17:37
bechamelpaepke: not if a thread is still running17:43
bechamelC-z + kill %1 does the trick17:43
paepkebechamel, ok. i know how i detach the console.17:43
bechamelor shorter: C-\17:43
paepkebechamel, i just started the trytond and wanted to kill it immediatelly17:44
paepkebechamel, i know it worked on trytond 1.6. Just wondering what happened.17:44
-!- gremly(~gremly@ has joined #tryton17:44
bechamelpaepke: I'm working with 1.617:46
bechamelpaepke: actually even with the trunk, a  C-c just after launching it kill the process instantly17:47
paepkebechamel, weird....17:48
paepkebechamel, ok. than it looks like a local issue to me :-)17:49
bechamelpaepke: maybe17:51
-!- paepke( has left #tryton17:57
-!- rednul_( has joined #tryton18:06
-!- enlightx( has joined #tryton18:19
-!- paepke( has joined #tryton18:22
phellerpaepke: ctrl-c with the server works in trunk...18:44
paepkepheller, just tried from my workstation. here it works. seems like a shell-issue.18:44
paepkepheller, apple, you know ;-)18:45
phellerpaepke: are you running xml-rpc?  ... the way it was in 1.6, the listener was synchronous, and therefore the listen wouldn't return.  That is what was fixed in trunk.18:46
paepkepheller, yes, running xml-rpc.18:46
phellerpaepke: if you can test it with trunk on both your systems, please let me know if you still have a problem so I can troubleshoot.18:46
paepkepheller, cause of a project involved proteus (great stuff) i'm using xml-rpc with trunk. thats what i have not used before.18:47
phellerah, I see18:47
phelleryes, proteus is very useful!18:48
paepkepheller, update. i can verify i cannot use ctrl+c over a ssh-connection when running trytond interactive on another machine.18:49
paepkepheller, on local machine running trytond i can kill it with ctrl+c18:49
paepkeboth running with xml-rpc18:49
phellerpaepke: and it is trunk running on the far end machine?18:53
udonopaepke: maybe your ssh client mask/catch/drop ctrl-c?18:55
udonopaepke: can you send SIGTERM from your ssh-client to the session?18:55
paepkepheller, it was 2 changesets beyond18:56
phellerpaepke: hmmh, then you should have the fixes for the xmlrpc listener18:56
phellercan you disable all but netrpc, and see if ctrl-c then works?18:57
paepkepheller, already done: it works when only using net-rpc18:57
paepkenow i'm on trunk.18:57
paepkepheller, if I enable xml-rpc i have this situation18:58
phellerpaepke: ok, one moment18:58
paepkeudono, good point. but it looks like an xml-rpc issue18:58
phellerpaepke: you can verify ctrl-c by doing something like "/bin/sleep 20", then hit ctrl-c.18:59
phellerpaepke: but, if you can ctrl-c with only netrpc, then this is working18:59
-!- pheller_(~pheller@2002:ad30:d8c3:0:217:f2ff:fe04:afe) has joined #tryton18:59
paepkepheller, works with ctrl-c18:59
paepkepheller even with ctrl-c18:59
paepkepheller even with sleep..19:00
phellerpaepke: yep, ok19:00
phellerpaepke: I'm looking at the source tree for the next test...19:00
paepkeudono, i can only kill it with SIGKILL19:00
paepkeusing "kill -9 <processid>"19:01
phellerpaepke: can you verify that protocols/ "XMLRPCDaemon" inherits from "daemon" ?19:01
udonopaepke: SIGKILL could indicate open transactions19:02
phellerpaepke: yes, good point from udono..... are you sure there aren't any transactions going on when you hit ctrl-c?  the thread.join() will block until they are all done19:02
paepkeudono, just starting the client. and killing imediatelly after "waiting for connections"19:03
paepkepheller, udono its on my dev machine.19:03
paepkepheller, udono how can i verify if there are open transactions? on database level? would be pretty cool to have a "trytontop"19:04
udonopaepke: so you need SIGKILL to kill the client?19:04
paepkeudono, to kill the server.19:04
paepkenot the client19:04
udonopaepke: k19:04
phellerpaepke: netstat is probably the easiest way to see open connections to trytond19:04
paepkepheller, by looking at the sourcecode  "XMLRPCDaemon" inherits from "daemon"19:06
paepkepheller, btw: not using the secure won19:06
phellerpaepke: ok, good.  I'm not quite sure what is happening for you then....  I will need to read the documentation for SocketServer, and then see if I can think of an idea to try19:06
paepkepheller, udono. you know. its not that important bug. but it looks like a bug and its anoying.19:07
phellerpaepke: it's important to me :-)19:07
phellerpaepke: we must be able to trust that SIGINT will shut the server down gracefully, or people will use SIGKILL, and wonder why there is garbage in their database  (partially completed things, etc)19:08
udonopheller: hopefully postgres will never store an unfinished transaction...19:09
phellerudono: per acid, it shouldn't.  And I really don't think it will.  But we should really shut things down as gracefully as possible.19:09
udonopheller: correct19:10
paepkepheller: ack. i'm using SIGKILL :-)19:10
paepkepheller my dev machine is debian lenny, my local workstation is debian squeeze. maybe a versioning problem?19:11
paepkedev:2.5.2 vs. workstation: 2.6.619:12
phellerpaepke: anything is possible.... there are definitely some big issues in SocketServer between 2.4 and 2.619:12
phellerpaepke: ah, so ctrl-c isn't working on 2.5.2, but it is on 2.6.6 ?19:13
paepkepheller, yes youre right19:16
-!- enlightx( has joined #tryton19:29
lem0nawhat about killall -9 trytond?19:35
paepkelem0na, yes, that works. well. its running "python bin/trytond"19:39
paepkelem0na, i can kill with SIGKILL (9)19:40
-!- cedk(~ced@gentoo/developer/cedk) has joined #tryton19:47
paepkeok, i'm filing a bug about this xml-rpc situation19:49
paepkecedk, i'm missing a version information on roundup. should i put it somewhere in the tags?19:52
udonopaepke: tag editing seems no longer avail.19:54
udonopaepke: just put the information into the description or the subject19:55
paepkeudono, done. just for the log:
phellerpaepke: I've summarized our troubleshooting in the bug too19:59
yangoonpaepke: can you switch to python 2.5 on your workstation?20:01
yangoonor 2.6 on dev machine20:01
paepkeyangoon, 2.5 on workstation.20:02
paepkeone moment20:02
yangoonwhatever erions are installed, you can switch easily by changing the link to the appropraite version20:02
yangoonor by doing something like python2.5 ..../trytond20:03
paepkeyangoon, python2.5 on workstation. cannot kill by ctrl+c20:04
paepkeyangoon, and verified again: python2.6 works.20:05
yangoonpaepke: ok, then problem is narrowed to python version20:05
paepkeyangoon, yes, good point to test it on same machine20:05
yangoonhave to leave now, cu later20:05
paepkeyangoon, cya20:05
-!- paepke( has joined #tryton20:27
phellerAnyone have an example of how to add multiple criteria to an ir.default clause?20:29
phellerfor instance, I want to set city=Belmont when zip=02478 and country=233 ....20:29
phellerI tried simple and between them, as well as pyson syntax, but neither seems to work.  perhaps the backend code can't yet handle this20:30
-!- zodman(~andres-va@ has joined #tryton20:36
cedkpheller: it is simply field_name=value20:39
cedkpheller: but it doesn't work for many clause20:39
phellerso I cannot do "zip=02478 and country=233"... ?20:39
cedkpheller: no20:40
phellercedk: sounds like a feature request is coming ;-)20:40
cedkpheller: in fact I think we should drop this functionnality20:41
phellercedk: the defaults ?20:41
cedkpheller: the change_default20:41
-!- irclog( has joined #tryton20:45
cedkpheller: I think it can be handle with on_change20:46
cedkI always thought that if we can drop a functionnality, we should20:48
phellercedk: I think rather, it should be "if we can simplify the code while keeping functionality, we should"20:49
cedkand change_default is only used in one place (party.address)20:49
cedkpheller: no we can drop functionnality if it is not used and can be replace by an other one20:50
phellercedk: we are saying the same thing, I think.... just with different words :-)20:50
phellercedk: I had another idea of how this could be done, and keep the performance....  how about the ability to have two fields refer to different fields in the same related record?20:51
phellercedk: so we could make a model for city/state/zip20:52
phellercedk: change one of them in party.address, and the others change to match20:52
-!- enlightx( has joined #tryton20:53
phellercedk: nice on the settimeout -- I will test on Mac later.20:53
cedkpheller: just write on_change method that do the job21:11
-!- mr_amit(~amit@ has joined #tryton21:14
phellercedk: yes, this is probably better... I don't think there are many places where linked fields to different fields of the same record are needed....  but do you think a city/state/zip/country model is a good idea?21:16
phellercedk: for fast entry in a small business, I think free text is faster for city....21:16
phellercedk: but, referential integrity is good.....21:16
cedkpheller: yes21:17
cedkpheller: It can extend the current country/state21:17
phellercedk: ok, good idea.  I will look at this after 1.8 is released.21:17
-!- bechamel( has joined #tryton21:19
-!- ikks(~ikks@ has joined #tryton21:34
-!- enlightx( has joined #tryton21:36
paepkecedk, short question. I had not checked it out currently, its enough to have a yes or no: Is it possible to get the xml-data for rendering the client-forms via xml-rpc?21:37
cedkpaepke: yes21:40
paepkecedk, great. thx.21:40
cedkpaepke: you must call fields_view_get21:40
paepkecedk, ok. looks good.21:43
lem0nawhere i can read what is aged balance (i am not familiar with financial terms)21:47
-!- zodman(~andres-va@foresight/developer/zodman) has joined #tryton21:53
cedklem0na: it is a report that show you who will owe money21:55
cedklem0na: how much and when21:55
lem0nacedk: something like chronologicaly (like timestamp)?21:56
cedklem0na: no more in 30 days, 60 days etc.21:57
lem0nacedk: thaks - will search for info in bulgaria for tje exact term21:57
cedklem0na: generaly it is not a legal report21:58
lem0nacedk: thanks21:58
cedklem0na: but it is used by commercials to inform customer about payment21:58
lem0nacerd: looks that i have to learn some finance :)21:59
udonocedk: hi, do you have an idea for
cedkudono: from where comes customer_tax ?22:33
cedkudono: ok understood22:37
cedkudono: you can not add the same record to two modellist22:38
udonocedk: yes, I need to instanciate them again22:38
udonocedk: Is this intended?22:39
cedkudono: yes22:39
udonocedk: ok. Why?22:39
cedkudono: because when a record is linked by a one2many or many2many, there is some on_change etc. that can occur22:43
cedkudono: so you must handle those change in separate instance to be sure to not have side effect22:44
cedklike infinite loop22:44
cedkudono: but perhaps proteus could create it for you22:45
cedkI don't know yet22:45
phellerhow would I use proteus to simply get the xml for a form view of an existing party?22:55
cedkpheller: call fields_view_get22:59
phellercedk: against which object?  assuming I do users = User.find() (where User is res.user model), do I just do users.fields_view_get() ?23:03
yangooncedk: hi23:03
yangooncedk: links on #tryton-commit not working : 00manifest.i@e9c01c409375: no match found23:03
phellercedk: I see that returns a list, but if I then do users[1].fields_view_get(), it claims res.user object has no attribute 'fields_view_get'.23:04
cedkpheller: against User23:07
phellercedk: ok, so then I get a different error:23:08
cedkyangoon: I had to rollback Bertrand commits23:08
phellercedk: xmlrpclib.Fault: <Fault 1: 'pop from empty list\23:08
cedkpheller: you must at least pass the context23:08
yangooncedk: ah, wondered already of disappearing commits23:09
paepkecedk, pheller and I am different people ;-)23:09
paepkepheller:  weird that you had asked the same question than i before.23:09
udonocedk: ok, thanks. I think we can improve the api of proteus, when we have some more example code.23:10
phellerpaepke: ok, I see now.  yes, I saw you asked earlier paepke, I was only wondering why it wasn't working for me ;-)23:10
-!- ikks(~ikks@ has joined #tryton23:24

Generated by 2.11.0 by Marius Gedminas - find it at!