IDempiere/FullMeeting20130227
⇐ Table of Contents | Full Meeting Minutes | Full Meeting 2013-02-27
CarlosRuiz: Good Morning
nmicoud: Bonjour
fcojvanninir: hello
JanThielemann: hi
Dumer: Salem
fcojvanninir: I had a question about the adempiere trees and the typical way of representing trees *
fcojvanninir: oops... that enter went too soon..
tbayen: Daarestiet
fcojvanninir: anyways, in the product category window you have a reference to parent product and I remember carlos saying in a lecturer that this broke the typical way that adempiere did trees.
fcojvanninir: Then I saw the way adempiere did tree and I under why it broke that norm.
fcojvanninir: How difficult would it be for you to reconcile both approaches. In other words, if a table has a reference to itself to signify the parent, that this would update the table were adempiere kept its corresponding tree structure.
fcojvanninir: Forming the tree for a table with many records can be quite a laborious task with the drag and drop.
nmicoud: IMHO, you can manage multiple trees for a same table (and use them through reporting hierarchy). That's why it shouldn't be possible to use Parent_ID column to determine the node parent (as a product, or any other record) could have a different parent per tree
Dumer: When will the iDempiere 1.0 release?
JanThielemann: are you sure you have to use a link to the parent in the same table to build tree structure? i thought thats what the treenode tables are for
fcojvanninir: Yes yes.. what i meant to say was that specifying the parent node updated the treenode tables
nmicoud: but what if you have 2 product trees ? both will be updated ?
CarlosRuiz: you would need to have a way to indicate "main prod cat tree" (as in client info tab) - and updating the tree from a prod cat model validator sounds easy - about the opposite I'm not sure - updating the parent column if the tree is moved doesn't sound so easy
CarlosRuiz: cos AFAIR the tree tables are updated with direct SQL
CarlosRuiz: nmicoud, what is the status of IDEMPIERE-528 Determine Zoom window depending on record
CarlosRuiz: is it ready for peer review?
nmicoud: in the AdempiereTreeModel class (the one with a bug inside for swing :)) => idempiere 531
nmicoud: Carlos : yes ; you could also review tickets 585 and 475 which are quite simple and ready to be integrated
nmicoud: Another thing : for ticket 90 and 622, there is a need for adding a button next to the table/table dir field. I have no idea for implementing it. Could it be done by someone else ?
Dumer: When will the iDempiere 1.0 release?
Terra_777: I also interested in the question about the 1.0 release
CarlosRuiz: our thinking is to make a freeze during march with the goal of a first release on april - that's the plan, be aware reality can lead us to something different :-)
CarlosRuiz: nmicoud, IDEMPIERE-90 works with right click?
Terra_777: More questions I do not haveMore questions I do not have =)
nmicoud: in webUI, yes ; but in swing mode, you have a padlock image, which is more user friendly
tbayen: k-man, are you still there?
CarlosRuiz: ok, sounds like presentation issue if you have another working way to do that - not high priority
CarlosRuiz: nmicoud, about IDEMPIERE-585 Accouting tabs should have InsertRecord = N
tbayen: I like the padlock.
CarlosRuiz: :-) I just realized that isAcctTab is called isInfoTab :-D
nmicoud: yes, that's... logical :))
CarlosRuiz: the patch doesn't seem right
nmicoud: why ?
CarlosRuiz: there are 44 active acct tabs - i.e. GL Journal
nmicoud: oops
nmicoud: so it should be like TableName ends with '_Acct' => IsInsertRecord= N
CarlosRuiz: I'm not sure of a second implication ... let me check
nmicoud: sure
CarlosRuiz: --> testing multiple acct schemas ...
CarlosRuiz: seems like I found a bug - delete button is enabled for M_Product_Acct but the table is not marked as "Records deletable"
nmicoud: in web ?
CarlosRuiz: yep
nmicoud: not a bug ; it has always like this, no ?
nmicoud: and when you click on the button, you have a msg "this record can't be deleted, deactivate it" ?
CarlosRuiz: I think the delete and delete selection button must be inactivated in such case
CarlosRuiz: is the same on swing?
nmicoud: absolutely
nmicoud: yes
CarlosRuiz: ok Nicolas - maybe a migration script instead of forcing the rule on MTab?
nmicoud: what if new tabs are created ?
nmicoud: the "TableName ends with '_Acct' => IsInsertRecord= N" doesn't sound good for you ?
CarlosRuiz: seems like too many exceptions for the rule
nmicoud: ok,
CarlosRuiz: this is the update
CarlosRuiz: update ad_tab set isinsertrecord='N' where isinfotab='Y' and isinsertrecord='Y' and (select tablename from ad_table tb where tb.ad_table_id=ad_tab.ad_table_id) like '%_Acct' and (select tablename from ad_table tb where tb.ad_table_id=ad_tab.ad_table_id) not in ('Fact_Acct', 'A_Asset_Acct', 'A_FundingMode_Acct', 'A_Asset_Group_Acct')
CarlosRuiz: 14 tabs - assets require inserting records as I see
nmicoud: is it normal ?
CarlosRuiz: I think so
nmicoud: not 'standard' ; those record should be created automatically (but not aware of Asser module)
CarlosRuiz: yes - what I saw in those Assets tabs is that there are more information to fill - not just accounts
Dumer: Has anyone done any load testing ADempiere?
nmicoud: is it a record per acctSchema ?
CarlosRuiz: yep
nmicoud: so, we should not have multiple record for a schema, no ?
CarlosRuiz: yep
tbayen: Dumer, there are some bigger installations. I myself have a little bit experience with importing/exporting bigger data tables. What exactly do you want to test?
nmicoud: so, no exception anymore ?
nmicoud: as you must not enter manually data in Fact_Acct window
CarlosRuiz: don't understand - the idea is that A_Asset_Group_Acct allows insertion
CarlosRuiz: Fact_Acct is a different kind - I guess it must be already read-only - but pointing is a different scenario
JanThielemann: Dumer, tbayen: I have tested Adempieres trees with 1000000 entries and they get realy slow/unuseable. fortunateley we made a lazy loading tree contribution but it still has to be implemented in idempiere
nmicoud: A_Asset_Group_Acct have the same structure as A_Asset_Acct ? and there should also be one record per AcctSchema ?
CarlosRuiz: don't know the details of fixed assets - but yes - I think must have one record per schema
nmicoud: so these tables shouldn't be treated as exceptions in your script, right ?
Dumer: Whether in ADempiere simultaneously 100,000 users?
CarlosRuiz: yes - they are - I think maybe is better to do the migration script manually for the 14 tables - in case somebody has a table ending in _Acct somewhere
JanThielemann: Dumer: no, we don't have the capacity to test this, maybe someone else? this sounds interesting
JanThielemann: but i think it could be possible since the clients connect directly to the database
nmicoud: I don't see a scenario where you have to enter multiple record for a single schema in those table. But, it's not very important, let's make a migration script, that should be good enough
CarlosRuiz: nmicoud, those asset acct tables are not prefilled
CarlosRuiz: as I saw you need to go there and fill them - not like the others
nmicoud: maybe an initialization process, or filling them with default could be a solution ?
CarlosRuiz: don't know the fixed asset details to answer that :-(
nmicoud: That's Edwin who did the last version ot asset ?
CarlosRuiz: yep - or red1 has experience with that also
nmicoud: maybe we could update the ticket (there is no emergency) and ask them what could be done ?
CarlosRuiz: sounds like a separate ticket
CarlosRuiz: this one could be solved changing the 14 tables
CarlosRuiz: tabs
nmicoud: yes, but i think we should arrive at the end to prevent creating records in those table ?
CarlosRuiz: or you can control duplicates with a unique index - checked one of those tables and it doesn't have unique index for that
tbayen: Dumer, this number should be no problem. The question is how many transactions generate them and how? If the all give you one order a week and you have a big callcenter with agents working on swing clients this will become quite a big load. I plan for myself a business with up to 50000 invoices per year (250 per day). My tests said this will work well without special things like separating databases, replication, traffic shaping
tbayen: etc. But until now It does not work in the real world.
nmicoud: Yes, but that sounds like the activated button on table where deleting record is not allowed.
nmicoud: That should not arrive, nope ?
CarlosRuiz: it depends - seems like those _acct tables are not following the same idea of the other _acct tables
CarlosRuiz: nmicoud, testing here IsShortList for first time :-)
nmicoud: ok, so : 1 - let make a migration script to put things right ; 2 - ask red1 and/or edwin about initializing asset ?
CarlosRuiz: it shows >Charge< and by default is "short listed"
nmicoud: you'll love it ;)
CarlosRuiz: just for lists - not for search windows - I also added to product and doesn't work
nmicoud: yes, those '<' and '>' are not really nice ; that's why i think an image would be better (you can try on swing)
nmicoud: yep
CarlosRuiz: testing on swing ....
nmicoud: you have more criteria on search window, but that could also be added, as they are using table (and not hardcoded anymore)
CarlosRuiz: swing not starting here :-(
Dumer: tbayen, for example, entering 56 million invoices?
Dumer: per year
nmicoud: you broke swing ? i've noticed that sometimes swing don't launch and i need to launch webUI before. Or sometime a restart of Eclipse is needed
CarlosRuiz: ah - a couple of bundles required were not marked
CarlosRuiz: found a bug :-) messages are not translated
CarlosRuiz: ShortListOnlyShortListItems and ShortListAllItems
nmicoud: you have the correct migration script ? i don't notice that
CarlosRuiz: found another bug
CarlosRuiz: I do "new" on sales order line
CarlosRuiz: typical :-) now is working fine when I tried to explain the problem
nmicoud: héhé
CarlosRuiz: does it remember your last selection of the padlock?
nmicoud: i don't think so
nmicoud: it should always show the short list
CarlosRuiz: on zk is doing that - I close sales order window and when reopened it shows the charge short list
nmicoud: yes
CarlosRuiz: but on swing is behaving different - I close sales order window and when reopened it remembers the last selection of the padlock (that's ok too)
nmicoud: not good, we should have same behaviour
nmicoud: and what do you think of the padlock image ? better than the < and > ?
CarlosRuiz: more accessible - without needing the right click
CarlosRuiz: the right click will be slow and difficult on tablets
nmicoud: yes
nmicoud: but actually, table fields don't allow images. And honestly, i'm not the guy who will do this. Can it be done by an expert ? then i could finish the job and add the padlock image ; that shouldn't be the more difficult
CarlosRuiz: I closed my swing client - and now it refuses to start again :-(
tbayen: CarlosRuiz, short list items messages are translatet in german translation. Who is maintaining the columbian trl? ;-)
CarlosRuiz: no, I mean in english - they appear as ShortListOnlyShortListItems and ShortListAllItems
CarlosRuiz: the tooltip of the padlock
nmicoud: that's weird ; did you have the msg in your AD_MEssage table ?
tbayen: A frech developer and a german tester are not good for english translations.
tbayen: french
CarlosRuiz: VLookup line 1705
CarlosRuiz: message doesnt't exist
nmicoud: :-/ do you have the good version ?
CarlosRuiz: I have the version on the migration scripts - and I see they have different values
nmicoud: tbayen, you have this problem ?
tbayen: mom... have to test on another machine
CarlosRuiz: __________________________________
CarlosRuiz: diff -r a308056929e6 org.adempiere.ui.swing/src/org/compiere/grid/ed/VLookup.java
CarlosRuiz: --- a/org.adempiere.ui.swing/src/org/compiere/grid/ed/VLookup.java Tue Feb 26 20:39:16 2013 -0500
CarlosRuiz: +++ b/org.adempiere.ui.swing/src/org/compiere/grid/ed/VLookup.java Wed Feb 27 09:38:01 2013 -0500
CarlosRuiz: @@ -1702,12 +1702,12 @@
CarlosRuiz: refresh();
CarlosRuiz: m_lookup.fillComboBox(isMandatory(), true, true, false, isComboShortList());
CarlosRuiz: m_buttonSL.setIcon(Env.getImageIcon("LockX16.gif"));
CarlosRuiz: - m_buttonSL.setToolTipText(Msg.getMsg(Env.getCtx(), "ShortListOnlyShortListItems"));
CarlosRuiz: + m_buttonSL.setToolTipText(Msg.getMsg(Env.getCtx(), "ShortListShortListItems"));
CarlosRuiz: }else{
CarlosRuiz: onlyShortListItems = true;
CarlosRuiz: m_lookup.setShortList(false);
CarlosRuiz: m_buttonSL.setIcon(Env.getImageIcon("Lock16.gif"));
CarlosRuiz: - m_buttonSL.setToolTipText(Msg.getMsg(Env.getCtx(), "ShortListAllItems"));
CarlosRuiz: + m_buttonSL.setToolTipText(Msg.getMsg(Env.getCtx(), "ShortListListAllItems"));
CarlosRuiz: }
CarlosRuiz:
CarlosRuiz: actionRefresh();
CarlosRuiz: __________________________________
CarlosRuiz: that must fix it - but need swing team to test - my swing refuses totally to start here :-(
tbayen: xchat opens balloons for every line in the chat. I can't work if you open so much lines... :-) Haha
nmicoud: ok, we will investigate ; thomas, when you have time, could you please test on a 'fresh' install ?
CarlosRuiz: that could be called spam patch :-D
tbayen: I did a fresh install for my production system the last days. Much other work to do. I hoped to find time at the weekend to sort the swing issues and create a new working version. I will test it then. I would be nice if you could patch before what you already have yourself. ;-)
hengsin: why the padlock image for a
nmicoud: ok, i will review it
hengsin: why the padlock image for a "shorten list" ? isn't some sort of icon that show it is filter or there's more items available more appropriate ?
CarlosRuiz: nmicoud - the patch is synchronizing the value of message with the inserted on 2013012430719_IDEMPIERE-90.sql
nmicoud: i use the padlock as it was the more appropriate from available image. But we could take a new one
CarlosRuiz: good idea hengsin - maybe a funnel image
nmicoud: yep
CarlosRuiz: http://www.iconfinder.com/search/?q=filter
tbayen: I just testet the tooltips of the short list items on a installation about 10 days old. It seems to work well. But I will test on a fresh install at the weekend.
tbayen: good idea from hengsin. I never thought about the icon.
CarlosRuiz: nmicoud, about IDEMPIERE-528 - there is no migration script - right? just the provided 2pack
nmicoud: yes i think (it's been a long time, don't remember well, but don't find a trace of migration script)
CarlosRuiz: ok - will try to do packing with migration script enabled and then fix the migration script
nmicoud: ok
hengsin: background: url(images/Lock16.png); background-repeat: no-repeat; padding-left: 18px;
hengsin: this css will give you a padlock image to the left of the label
hengsin: plus the right click menu, I guess it is good enough for now
nmicoud: sorry, trillian crashes...
nmicoud: hengsin, using you code for css will show the button, but can i interact with ?
hengsin: no
hengsin: you can make the label clickable but that will conflict with the existing zoom implementation
hengsin: one option will be to show a popup menu when user click on the label, i.e to zoom or to change the "shortlist" setting.
nmicoud: i have no idea how much time it take to have the ability to add an image to table field , but it could be nicer, as there is no emergency for this ticket
nmicoud: and this feature (image for table field) could also be used for ticket 622
hengsin: hmm ... actually it could be possible using some client side coding ( js ). not sure but that's quite a challenging path to take.
nmicoud: Wouldn't be the same thing as the button used for search fields ?
nmicoud: or location ?
hengsin: the current component we used for table field doesn't allow that. will need to build a custom one if that's what you want. however, that would mean you need to test a few more places to ensure the layout of places not effected ( grid, some custom form, etc )
hengsin: search field is using a different component
hengsin: i means not from the same hierarychy ...
CarlosRuiz: could we add a button besides the list - like payment?
nmicoud: even if it take more time, wouldn't be a better solution ? and with the freeze period which is coming, there should be enough time to strack bugs ?
nmicoud: *track*
hengsin: yeah, should be better. anyway, I don't think we need to freeze small UI refinement like this until we are really close to release.
nmicoud: and can it be done after release 1.0 ?
tbayen: gtg, thanks for now.
hengsin: yeah, of course.
nmicoud: so, as there is no emergency, that could be done after. So we could have then a better solution, no ?
CarlosRuiz: nmicoud, the 2pack is on 361 format :-) and no zk implementation yet?
nmicoud: yes, it was done on 361. No zk for the moment
CarlosRuiz: I think I'll need to postpone the review for that one :-(
CarlosRuiz: gtg too guys - thanks a lot for the meeting
nmicoud: i could create a migration script for this one if you want ?
nmicoud: for idempiere
CarlosRuiz: yep - that will ease it a lot
nmicoud: ok, i'll do this
nmicoud: bye
CarlosRuiz: bye thanks
