IDempiere/FullMeeting20120919

From WikiQSS

Table of Contents | Full Meeting Minutes | Full Meeting 2012-09-19

CarlosRuiz: Morning
nmicoud: Bonjour
CarlosRuiz: any ticket suggested for peer review today?
nmicoud: i'd say 382 (did not have time to tork on) ; 293 (list of logged users would be helpful also)
CarlosRuiz: ah yes - I was going to ask if you had time for that one
nmicoud: nope.... i've just try for few minutes
nmicoud: too busy :(
nmicoud: i've started to add some code within cmd_new of APanel (we said that when pushing button, record should be marked as changed)... but did not succeed
nmicoud: which method would you use to doing it ?
CarlosRuiz: let me check ....
CarlosRuiz: checking GridTable.dataNew
nmicoud: the idea would be to customize it by adding some controls ?
CarlosRuiz: do you mean 382?
nmicoud: yes
CarlosRuiz: no - there must be some point where the record is marked as not updated
CarlosRuiz: first clue I found (but didn't work) on GridTable line 2503
CarlosRuiz: m_rowChanged = -1; // only changed in setValueAt
CarlosRuiz: but tried that approach and didn't work - need to check somewhere else
nmicoud: what about using something like getMandatory(rowDAta) which should check only parent column ?
CarlosRuiz: I think the problem is on GridTable.needSave
CarlosRuiz: it's explicitly ignoring "New unchanged records"
nmicoud: i see
CarlosRuiz: ok - got a patch - but I'm not sure now - it has some collateral effects
nmicoud: great! need to be tested
CarlosRuiz: Nicolas - after the patch - then there are undesirable navigation effects
CarlosRuiz: this is my test case
CarlosRuiz: defined preference for the mandatory fields on sales order
CarlosRuiz: so when I push New all fields are mandatory fields are filled
CarlosRuiz: I navigate to order lines - after the patch the order is saved
CarlosRuiz: the bad thing is that if I navigate back to order header - then an order line without product or charge is saved
CarlosRuiz: I like how it works when autocommit is set to false - but not when autocommit is true
CarlosRuiz: do you want to check it?
nmicoud: sure
CarlosRuiz: this is the patch
CarlosRuiz: diff -r 8c36d38a8f09 org.adempiere.base/src/org/compiere/model/GridTable.java
CarlosRuiz: --- a/org.adempiere.base/src/org/compiere/model/GridTable.java Mon Sep 17 22:12:48 2012 -0500
CarlosRuiz: +++ b/org.adempiere.base/src/org/compiere/model/GridTable.java Wed Sep 19 08:39:46 2012 -0500
CarlosRuiz: @@ -1151,7 +1151,7 @@
CarlosRuiz:
CarlosRuiz: // Indicate Change
CarlosRuiz: m_changed = changed;
CarlosRuiz: - if (!changed)
CarlosRuiz: + if (!changed && !m_inserting)
CarlosRuiz: m_rowChanged = -1;
CarlosRuiz: //if (changed)
CarlosRuiz: // fireDataStatusIEvent("", "");
CarlosRuiz: @@ -2500,7 +2500,7 @@
CarlosRuiz: }
CarlosRuiz: }
CarlosRuiz:
CarlosRuiz: - m_rowChanged = -1; // only changed in setValueAt
CarlosRuiz: + m_rowChanged = m_newRow; // IDEMPIERE-382
CarlosRuiz:
CarlosRuiz: // inform
CarlosRuiz: log.fine("Current=" + currentRow + ", New=" + m_newRow);
CarlosRuiz: for iDempiere
nmicoud: i've only 361 ; that should work ?
CarlosRuiz: yes - it applies fine on 361
nmicoud: ok, i'm testing
nmicoud: did not tested with preferences for fields, but seems ok (with autocommit enabled or disabled)
nmicoud: but surely need more tests
CarlosRuiz: it works - but I don't like the effect of saving that empty orderline
nmicoud: but did you force a charge or product on document ?
CarlosRuiz: no - not forced, that would solve the case for orders - but the same can happen on any window that has details
CarlosRuiz: and the detail has not mandatory fields
CarlosRuiz: or mandatorys are filled with defaults
nmicoud: that's not working for included tabs
CarlosRuiz: the actual included tabs approach was already dropped from iDempiere - we're moving to a different approach
nmicoud: even for swing ?
CarlosRuiz: haven't tested on swing
nmicoud: another problem : go to a window where there is no records ; click on New, then Undo ; you can go on lines tabs (tested on swing with shipment)
CarlosRuiz: yes - I don't like this solution of auto-saving anyways
CarlosRuiz: I guess you're pointing to a different approach is about forbidding navigation to a detail
nmicoud: that's it; if you want to go in tab with a level >0, all parent field must be filled
CarlosRuiz: nmicoud
CarlosRuiz: finally got a working thing on zk idempiere
nmicoud: good
nmicoud: only for web ?
CarlosRuiz: yes - trying to check if it works for 361 too
nmicoud: thanks
CarlosRuiz: you could try to follow similar approach on VTabbedPane
CarlosRuiz: it works :-)
CarlosRuiz: will upload patches for your review
nmicoud: ok will do that, thanks
nmicoud: you said that you drooped included tab from iDempiere ; what you planned to do will look like this : http://wtcindia.wordpress.com/2012/06/17/included-tabs-as-horizontal-tab-in-adempiere/ (see last picture) ?
CarlosRuiz: ready Nicolas - uploaded to http://jira.idempiere.com/browse/IDEMPIERE-382 with comments guiding you where to patch swing
CarlosRuiz: we checked the wtc approach - no, I think is not similar to that one
nmicoud: wtc=?
CarlosRuiz: wtcindia
CarlosRuiz: walkingtree
CarlosRuiz: :-)
nmicoud: ok
CarlosRuiz: will wait for your tests on 382 to commit
nmicoud: and this new approach will be available fo both swing and web ? or just web ?
CarlosRuiz: I think just web
nmicoud: i'll try to test it now, or probably tomorrow
nmicoud: :(
nmicoud: would'nt be possible to let included tabs for swing ? could be very useful
CarlosRuiz: well - we're still not sure about the outcome of this one - after we refine it on zk maybe somebody will develop or sponsor the swing port
nmicoud: ok
CarlosRuiz: Nicolas - thanks for the meeting - need to move to a different task
nmicoud: bye bye ; see u later