Mike Vincent (Vee2d2) asked what does setting prequery on the datasource do? Derek Neighbors (derek) explained this option prequeries on start up. It used to be mandatory for lookup boxes - otherwise the lookup boxes would be blank :) So if you had a parts database - with a vendor field and the vendor field had a vendor id - you had a combo box that showed valid vendors to pick from to populate that field. If you didnt prequery the vendor datasource - then that combo woudl be blank on form startup.

Stuart Quimby (ToyMan) said he was looking to pick up the thread again - a new investor in his business was looking for us to upgrade our accounting system. Mike Vincent (Vee2d2) said there had been some progress on GNUe Small Business - he was going to try and get at least something written for each thing this weekend and add a link to 'em from the main page.

James Thompson (jamest) said we need to release the tools again - work is ongoing in cvs but at a lesser pace. He believed that Jason Cater (jcater) was mainly focusing on a GNUe-based POS (Point of Sale) system at the moment, which he needed personally. Mike said this was looking pretty good - made me wanna hook up my bar code reader and start zapping things in the office when I played with it.

James said that Bajusz Tamás (btami) had been adding functionality to GNUe Forms, including a native Win32 user interface driver, and someone added activex support I see (something I never envisioned :). He knew several people have added killer features to cvs - btami's work in windows alone deserves a release - however a few weeks ago I think jason needed a few weeks to fix up designer - so we're in a holding pattern. However, he can't recall any changes in forms that require a new designer - so i guess nothing would stop a forms/common release if it was needed, with Designer following later.

Mike Vincent (Vee2d2) explained that a Stock Keeping Unit (SKU) was a way of identifying products for inventory - a 'part number' - typically, a SKU is a number which represents an exact item, as for example my SKU# S12030104 might be a Nike model#244995 golf shirt that is Black and Large size.

Jason Cater (jcater) said I should have G/L committed soon for GNUe Small Business - using the gnue schema definitions - it's pretty much just arias w/gnue forms screens.

Later, Chan Min Wai (dcmwai) said he had been searching all over the net for a "standard way" of work doing (ERP, Stock ...) and I found that... it is all about money ... Mike said that's pretty much what the product (or item as it's currently called) management module of GNUe Small Business is about.. setting up a line of products, being able to build intelligent SKUs from pieces of information (if that's what you want).

Mike Vincent (Vee2d2) asked whether the on-startup trigger was fired before the others. Seems all the focus related triggers and on-newrecord etc, all get fired prior to on-startup.. so I get this chicken and the egg thing happening where I setup global variables in on-startup they get changed in an on-change trigger that gets data from a source and I was going to try using a pre-focusout trigger to update the data's source to reflect any changes. James Thompson (jamest) thought to set it up so that the app can init the trigger system then have a flag that says ignore triggers until flag is set. - what we'd do is change the gnue-forms so that right before it fires on-startup it does a IntializeTriggerSystem() or some such thing. James said he'd take a look at it.

Mike Vincent (Vee2d2) found out the problem that while doing a createResultSet(...) to test for the presence of data before proceeding, in case of empty result set, a new record was created. Jason Cater (jcater) asked whether the datasource was bound to a block and Mike confirmed that. In this case Jason offered to use datasource.createResultSet() then if block.isEmpty(): foo checking. Mike didn't think that this would help because the problem was I would enter some things, do the check and *poof* I'm on a new record.. where I didnt want to be. He said another datablock creation had helped.

Derek Neighbors (revDeke) asked Jason Cater (jcater) is all the stuff you are checking into gnue-sb basically arias stuff? as it's starting to look a real mess, citing a few specific concerns - i guess i just get nervous with massive cvs changes and adds. Chan Min Wai (dcmwai) thought These change is just tempory. Later, Jason confirmed there is no long-term arias/ directory - it will be rm -rf'd shortly and in any case that's a repository, btw... it's not part of the gnue-sb cvs tree.

The next day, Derek and Jason discussed how to structure the GNUe SB CVS as the arias code was being ported. Jason explained that the gnue-sb/arias subdirectory was a seperate CVS check out. Derek said this was his preference, but figured it would be too much a pain :) Jason explained ] I did it because it wasn't really officially part of gnue-sb - and I just needed cvs somewhere. Derek agreed - as the goal is to move things into gnue-sb/ eventually.

Mike Vincent reported that the schema navigator blows up when I try to open the Tables node under gnue_sb now and I'm coming up empty.. I dropped the db and started fresh even.. ;) which uncovered something else in the schema. Jason Cater (jcater) said you are creating the GNUe Point of Sale (POS) tables - those probably aren't ready. Mike said this was working if he dropped all the tables and the related sequences, indicies.

Later, he deduced It's a popy bug. I guess popy isnt return the oid as an int - switched back to psycopg and everything worked fine again. Jason said he had hit several issues w/popy in debian sid so had to switch back to psycopg. Mike noted that when I was trying to figure out what was happening I went into common's datasources/drivers/postgresql/Base/Driver.py and commented out the block of code ~188 that creates the list and instead printed what was getting returned - it appeared to be returning string rather than numeric (object ID) results. Jason said this matched what he had been seeing - it doesn't return native types. popy was wrapping native types for some reason with it's own classes - but they hadn't implemented all the __whatever__ methods - so some of our stuff was bombing.