*** hieplq has joined #idempiere | 01:31 | |
*** is-mw2 has quit IRC | 02:04 | |
*** is-mw has joined #idempiere | 02:05 | |
*** CarlosRuiz has quit IRC | 02:19 | |
*** hieplq2 has quit IRC | 03:06 | |
*** shmutalov has quit IRC | 06:07 | |
*** norbertbede has joined #idempiere | 07:19 | |
*** norbertbede has quit IRC | 07:33 | |
*** norbertbede has joined #idempiere | 07:47 | |
JanThielemann | good morning @ all | 07:50 |
---|---|---|
*** nmicoud has joined #idempiere | 07:51 | |
*** a42niem has joined #idempiere | 08:09 | |
*** is-mw has quit IRC | 08:41 | |
*** is-mw has joined #idempiere | 09:00 | |
*** shmutalov has joined #idempiere | 09:10 | |
shmutalov | ) | 09:11 |
JanThielemann | today i learned when i can use MyClass.this effectively | 09:14 |
JanThielemann | sometimes i feel realy stupid because of not knowing thinks like this years before | 09:17 |
JanThielemann | When the server is started, a new session with ad_user_id 0 is created | 09:44 |
nmicoud | hi jan, related to a scheduler ? | 10:16 |
*** KermitTheFragger has joined #idempiere | 10:24 | |
*** norbertbede has quit IRC | 11:28 | |
*** norbertbede has joined #idempiere | 11:28 | |
*** norbertbede has quit IRC | 11:35 | |
*** hieplq has quit IRC | 12:53 | |
*** CarlosRuiz has joined #idempiere | 13:45 | |
*** ChanServ sets mode: +o CarlosRuiz | 13:45 | |
*** norbertbede has joined #idempiere | 13:51 | |
*** Adam_ has joined #idempiere | 13:51 | |
Adam_ | hi | 13:51 |
Adam_ | I've got a question about idempiere -> is there a simple way to change primary key column length? For now all of them are numeric(10,0). Is there a simple way to change them to numeric(18,0)? | 13:53 |
JanThielemann | why would you do that? | 13:53 |
Adam_ | Beacouse I have to | 13:53 |
JanThielemann | :D | 13:53 |
Adam_ | :) | 13:53 |
JanThielemann | so... why do you have to change the lenght? | 13:54 |
JanThielemann | i don't think that you want to spend some hours just for fun | 13:54 |
Adam_ | No, it's no fun at all :D Let's say the reason is not relevant, but there is no other way I can do something than changing the length od those columns. | 13:55 |
Adam_ | Is there a way to do so? | 13:56 |
JanThielemann | yes but you will not like it :) | 13:56 |
Adam_ | SQL? | 13:57 |
JanThielemann | yes | 13:57 |
JanThielemann | and make sure you don't forget any of the 840 tables | 13:57 |
JanThielemann | any you may have to drop and recreate some views during the process :) | 13:57 |
CarlosRuiz | I think you're restricted by the int type in all java code - max int is 2147483647 | 13:57 |
Adam_ | actually I've tried that -> I have pretty nice ALTER statement to change all the columns I want, but there is a problem with views | 13:58 |
Adam_ | There is BitInteger in Java ;) | 13:59 |
Adam_ | BigInteger | 13:59 |
CarlosRuiz | you would need to change all idempiere code to manage this new type for the ids | 13:59 |
Adam_ | You're saying that iDempiere want't handle well that kind of ID? | 14:00 |
Adam_ | damn | 14:00 |
CarlosRuiz | I'm saying the effort can be huge - so - back to the JanThielemann question - review the need - probably there is another easier way to achieve it | 14:00 |
JanThielemann | idempiere uses int for id columns | 14:00 |
Adam_ | OK, so that out of the table than | 14:01 |
CarlosRuiz | we have also thought about the possibility of dropping all the IDs and use the UUIDs instead - but again, the effort seems big | 14:02 |
Adam_ | My need in synchronization with POS software on mobile devices. I'm using SymmetricDS to synchronize some tables from iDempiere. The problem is that for orders I need to synchronize c_order and c_orderline which links to c_order and I need to find a way of synchronizing PK,FK without conflicts. | 14:03 |
Adam_ | UUID would be great ;) | 14:04 |
Adam_ | SymmetricDS is a database synchronization software. | 14:04 |
*** norbertbede has quit IRC | 14:05 | |
CarlosRuiz | maybe you can use the UUID for synchronization - consider UUID as an alternate primary key | 14:06 |
*** norbertbede has joined #idempiere | 14:06 | |
Adam_ | For synchronization to POS I could use that, but there will be problem with synchronization back if I would like to create order records on POS (which will be without IDs) | 14:07 |
Adam_ | Main problem is that POS needs to create order records | 14:08 |
Adam_ | with orderlist | 14:08 |
Adam_ | And to synchronize it back to iDempiere I need to have c_order_id in c_order and c_orderline already set | 14:09 |
Adam_ | My idea was to concatenate it from some POS_ID + POS_SEQUENCE on POS so they would be unique, but length of 10 would be to short | 14:10 |
*** hieplq has joined #idempiere | 14:38 | |
Adam_ | thx for help and goodbye :) | 14:45 |
*** Adam_ has quit IRC | 14:45 | |
*** allgood has joined #idempiere | 15:18 | |
*** allgood has left #idempiere | 15:19 | |
norbertbede | hi. all | 15:31 |
norbertbede | CarlosRuiz could you please commit this https://idempiere.atlassian.net/browse/IDEMPIERE-2469 | 15:32 |
norbertbede | its fixed by hiep we are deployed, and looks 99% ok | 15:32 |
norbertbede | but sometimes issueas appear. but i mean for further investigation we need to get in core. pls | 15:32 |
CarlosRuiz | done | 15:36 |
norbertbede | thanks well | 15:37 |
nmicoud | hi CarlosRuiz : on your last comment on https://idempiere.atlassian.net/browse/IDEMPIERE-868, you refer to a patch. Which one ? | 16:05 |
*** ocurieles has joined #idempiere | 16:05 | |
CarlosRuiz | patch_868_MColumn.patch | 16:07 |
nmicoud | ok | 16:08 |
nmicoud | looking at "my" MColumn class, there are comment about IDEMPIERE 868 and 965. Need to check to see what i really used (all modifications were done in 2013). Will give feedback asap | 16:11 |
CarlosRuiz | you know I don't use oracle - so don't know if that problem arises and which length of table is required then - another restriction imposed by that :-) | 16:12 |
nmicoud | :) | 16:12 |
nmicoud | my class seems a little bit different. Will upload a patch on the ticket. And ask for other Oracle users to check | 16:13 |
CarlosRuiz | oracle looks like windows 95 with such restriction :-) | 16:13 |
nmicoud | something like that :D | 16:14 |
CarlosRuiz | or DOS :-) | 16:14 |
nmicoud | :) | 16:14 |
CarlosRuiz | I remember when you just had 8 characters to name your files in a diskette | 16:14 |
nmicoud | yes - it was 20 years ago :d | 16:15 |
CarlosRuiz | nmicoud, are you still using swing? | 16:34 |
nmicoud | only in System (i'm the only one) | 16:34 |
nmicoud | customers only use web | 16:34 |
CarlosRuiz | ok - just checking that are a lot of swing related tickets - and if there is no interest maybe is better to close them and label as potential idea | 16:35 |
CarlosRuiz | I'm today triaging old tickets | 16:35 |
nmicoud | i think you can | 16:35 |
nmicoud | AFAIK, only tbayen and me were using it | 16:36 |
CarlosRuiz | why don't you use web for System? :-) | 16:36 |
CarlosRuiz | I'm now so used to all the new features of webui - that I see the swing client as boring | 16:37 |
nmicoud | web is slower | 16:38 |
CarlosRuiz | I don't feel that - maybe as I don't use swing I just got used to the speed of web | 16:38 |
nmicoud | i think | 16:39 |
nmicoud | sometimes i use both at same time, and web is soooo sloooow | 16:39 |
norbertbede | nmicoud if you will use webui for system i will help test you :) more people more experience :) | 16:55 |
norbertbede | we drop swing in our strategy | 16:55 |
nmicoud | one time, i try to create a whole thing in web : table/window/process/... and i was mad. I think i do things 2 or 3 x fast in swing | 16:57 |
norbertbede | then we must make better zk ;) | 16:57 |
nmicoud | :D | 16:57 |
hieplq | @nmicoud, you use a high speed program but let your user a verry lower speed program, how to you can understand his unhappy :D | 17:01 |
nmicoud | :) thay don't know the "power" of swing | 17:02 |
CarlosRuiz | I don't think swing is that fast - really - we could try to measure how much time you spend creating a table in swing and I in zk | 17:02 |
norbertbede | CarlosRuiz i mean its not only about time but comfort | 17:03 |
norbertbede | e.g. window/tabs navigation | 17:03 |
CarlosRuiz | same here - I feel more comfortable with zk - a lot more | 17:03 |
norbertbede | i prefer zk8 well | 17:04 |
norbertbede | hops zk7 ;) | 17:04 |
norbertbede | its about requirements too :) | 17:05 |
norbertbede | apropo, we have lot zk7 tickets ready to commit (A) | 17:05 |
CarlosRuiz | also - I just use local network for development - all my production and test servers are in the cloud - so swing is not an option there | 17:05 |
CarlosRuiz | yes - I have a big backlog on zk7 tickets | 17:06 |
nmicoud | production is in our cloud, so i can use swing, and i can garanty that it is faster than zk. A good example is to create an element and translate it immediately. | 17:07 |
nmicoud | but gtg now, will come back later | 17:07 |
CarlosRuiz | we can measure how much time you spent creating the element and translating vs me in zk :-) | 17:09 |
CarlosRuiz | let's do a race :-) | 17:09 |
norbertbede | anyway synch transl. is quite slow well | 17:12 |
CarlosRuiz | how slow? | 17:13 |
norbertbede | take in aws 5min on 8GB ram java instance + RDS 8GB PGSQL | 17:14 |
norbertbede | i know not too much resource but averagy of CPU is approx 25% only some processes are sloq | 17:15 |
norbertbede | w | 17:15 |
norbertbede | instance has SSD disks | 17:15 |
norbertbede | both | 17:15 |
CarlosRuiz | I noticed that when using the new postgres | 17:15 |
CarlosRuiz | is about memory | 17:15 |
CarlosRuiz | postgresql memory | 17:15 |
norbertbede | you mean doesnt fulfill minimum ? t waht you mean | 17:16 |
CarlosRuiz | I noticed this | 17:17 |
CarlosRuiz | in postgresql 8.4 the sync trl was very fast | 17:18 |
CarlosRuiz | seconds | 17:18 |
norbertbede | i see | 17:18 |
CarlosRuiz | now - when using postgresql 9.X default configuration | 17:18 |
CarlosRuiz | it takes too much time | 17:18 |
CarlosRuiz | and lot of CPU | 17:18 |
CarlosRuiz | there is one of the queries that postgresql cannot resolve properly | 17:18 |
CarlosRuiz | I change postgresql.conf to increase some memory | 17:18 |
CarlosRuiz | maybe shared_buffers or something like that | 17:19 |
CarlosRuiz | for production systems I use pgtune to help with that | 17:19 |
CarlosRuiz | and then the sync trl again takes seconds | 17:19 |
norbertbede | hm interesting | 17:19 |
norbertbede | im using pgbadger for analyse logs | 17:19 |
norbertbede | indexes etc | 17:19 |
norbertbede | thinking about pgtune usable for RDS | 17:20 |
norbertbede | but i mean not because RDS is service | 17:20 |
norbertbede | not linus | 17:20 |
norbertbede | linux | 17:20 |
norbertbede | found something here: http://www.davidmkerr.com/2013/11/tune-your-postgres-rds-instance-via.html | 17:20 |
norbertbede | ok.thanks for tip going to test :) | 17:21 |
CarlosRuiz | that looks good - as in RDS you don't have access to the postgresql.conf | 17:21 |
norbertbede | you have parameter groups - its same like postgressql.conf | 17:22 |
norbertbede | just create some parameter groups | 17:22 |
norbertbede | and you can switch on purpose | 17:23 |
norbertbede | e.g. 1 group = testing, 2nd production etc | 17:23 |
*** hieplq has quit IRC | 17:28 | |
*** KermitTheFragger has quit IRC | 17:36 | |
norbertbede | CarlosRuiz. Synch Translation after some tweaks took 6 sec instead 5min :-O | 18:02 |
norbertbede | thanks | 18:02 |
norbertbede | i followed http://www.davidmkerr.com/2013/11/tune-your-postgres-rds-instance-via.html | 18:03 |
CarlosRuiz | seems like a flaw in the default postgres config | 18:03 |
*** norbertbede has quit IRC | 18:46 | |
*** ocurieles has quit IRC | 19:10 | |
*** nmicoud has quit IRC | 19:25 | |
*** norbertbede has joined #idempiere | 20:03 | |
*** ocurieles has joined #idempiere | 20:30 | |
*** ocurieles has quit IRC | 21:53 | |
*** ocurieles has joined #idempiere | 21:54 | |
*** ocurieles has quit IRC | 21:58 | |
*** ocurieles has joined #idempiere | 22:37 | |
*** norbertbede has quit IRC | 23:15 | |
*** norbertbede has joined #idempiere | 23:15 | |
*** a42niem has quit IRC | 23:31 | |
*** ocurieles has quit IRC | 23:55 | |
*** ocurieles has joined #idempiere | 23:55 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!