GNUe Traffic #36 For 6�Jul�2002

By Peter Sullivan

"If I can't give derek a hard time then what fun is left in GNUe?" - "you mean, besides all the hot chicks? - oh, wait..."

Table Of Contents


This Cousin covers the three main mailing lists for the GNU Enterprise project, gnue ( , gnue-dev ( and gnue-announce ( . It also covers the #gnuenterprise IRC channel. A great deal of development discussion for this project goes on in IRC. You can find #gnuenterprise on, or you can review the logs ( . For more information about the GNU Enterprise project, see their home page at ( .

1. Documentation standards for GNUe

26�Jun�2002�-�30�Jun�2002�Archive Link: "[IRC] 27 Jun 2002"

People: Derek Neighbors,�Jason Cater,�Christian Selig,�Andrew Mitchell,�James Thompson

There were several queries about trying to compile some of the GNUe documentation from source using DocBook, which re-ignited the discussions from Issue�#32, Section�#6� (29�May�2002:�Documentation standards for GNUe) . James Thompson (jamest) said he had been unable to get the docbook tools to work. Derek Neighbors (dneighbo) noted that "the drawing program in staroffice is pretty good (er open office)" . Jason Cater (jcater) suggested "that's why we should do all docs in OpenOffice" , and ducked. However, Derek said he "wouldnt complain too loud really about openoffice docs" . Christian Selig (sledge_) asked "does open office run well on my p166 w/ 64mb ram?" . Jason said "definitely - albeit very, very slowly :)" . Andrew Mitchell (ajmitch) felt "the app itself is a bit heavy for me, but if you could produce docs with XSLT it'd be fine :)" . Derek said that he sometimes ran OpenOffice "as p233 64mb - start up is horridly slow - but after started it runs ok" . Christian suggested "why not use something that anyone can produce, anyone can read and anyone can understand? --> ascii *g*" Derek said "thats why im so pro docbook - is its just xml - which bsically is ascii - problem is no good editor (if you dont want to edit ascii (with xml tags)) AND the 'output generation tools' cause some folks to go delirious :) - er more delirious" . OpenOffice "does work aon about any platform reasonably well" but was resource-hungry and hence "not great on older hardware (but usable on it)" . James said that people without OpenOffice could just unzip it and edit the XML with any text editor. Derek said "i still say we take any format - but its likely my next few docs will be in openoffice" . Jason said he tended to use either a simple text editor or else "in something like OO or abiword - then transcribe into whatever format :) - so doing it in OO to begin with would r0x0r" .

Derek said "i like the ideals of docbook, but if people arent using it why bother" ? He noted that "at LWE there is a session on documentation using OS tools" - he was "hoping to find somoen has a docbook visual editor we just havent heard of" . Jason agreed - "I find it udderly amazing that ppl do all these docbook-based docs - and there's no good editors" .

Some days later ( , Derek said "um i vote open office offical documentation tool - i have been using oo drawing program at work to do some diagramming and its VERY VERY VERY easy to work with - blows visio/dia out of water for 'ease of use'" . James felt it was "too generic" . Derek said "i agree to a degree - if there is a way to get some custom elements it would be better - but still as it sits for simple diagramming its very nice i.e. EASY to use. I like dia but its very 'rough' i.e. its hard to make 'professional' looking polished documents. visio is well too expensive and well a bitch to use" .

2. Forms usability issues

26�Jun�2002�-�30�Jun�2002�Archive Link: "[IRC] 27 Jun 2002"

Topics: Forms

People: Derek Neighbors,�Christian Selig,�Reinhard M�ller,�Micah Yoder,�Jason Cater

Christian Selig (sledge_) asked how to query back a form filtered for a single value of a particular form. Derek Neighbors (dneighbo) said this could be done from within the form itself - "simple do f7 i thik (prepare query) - type USERINPUT% in the entry - hit f8 (i think(execute query) and you have all records matching that" You could put conditions (including SQL wildcards) into as many fields as you wanted to restrict the query. Christian said "now, this is *not* intuitive for my users :)" Derek said "i thought same thing - now after a few days my users are demanding other application have this functionality" .

Derek added "we will soon have 'search' boxes" . This would allow form designers to put a button next to a field - "you click that and get a search dialog" which you could search and select in, with the value automatically entered in the field when you exited the search dialogue box. Reinhard M�ller (reinhard) said that "my users like to find it very comfortable to have the system guess what they are wanting - if i enter 6890 there it looks up that's the zipcode of lustenau - and if i enter "berlin" there it knows that it can't be a zipcode then it shows a list of all cities starting with "berlin" - that's the "fastest" system wrt data entry" . Derek said "thats all trigger stuff - we have triggers in the smalpes TODAY that support your first example" . He added "certainly one could look into triggers that make zipcodes a dropwon and citsies adrop down - and based on what you pick in one the contents of the other is changed" .

Some days later ( , Micah Yoder (MicahY) asked whether it was possible to query back values in forms by ranges - "like select all records with id > 10 and < 20?" . Jason Cater (jcater) said "not yet" . Micah also suggested adding "a lot more visual indicaters in forms" to tell you when you were in prepare query mode. Jason said "there's the status bar" . Micah agreed, but felt "it's easy to miss though. :)" . Coloured backgrounds could be used to indicate mode, "and a red X or something beside a deleted record would be good.... (in a grid)" .

3. Converting applications to GNUe

26�Jun�2002�Archive Link: "[IRC] 27 Jun 2002"

Topics: DCL

People: Derek Neighbors,�Stuart Quimby

Stuart Quimby (ToyMan) asked about using GNUe to replace a multi-user database that was beginning to outgrow its application (Microsoft Access). Derek Neighbors (dneighbo) said "if i know your are drop dead serious adn ready to put real push into it im willing to put time into it" . The biggest problem was actually looking at Stuart's existing application as "its been over 6 months since i have had windows" . He would need access to Stuart's existing application, but not necessarily live data. "within a day i can probably have tables and data moved to postgres for you" . He recommended using DCL, "GNUe's project management tool" , to help "'manage' expectations" . He explained "think of it as glorified todo list - with groupings and workflow (notification) - also as you move to production it will be nice to have the bugtracking / complaint side of it :)" .

4. Using .grpc files to support remote procedures in different languagues

27�Jun�2002�Archive Link: "[IRC] 28 Jun 2002"

Topics: Application Server, Common

People: Perry Lorier,�Andrew Mitchell

Perry Lorier (Remosi) asked "does gnu-rpc use a common idl format? Interface Description Language" . He explained "to do RPC you need to know what the procedure *is* - corba does this by using .idl files" Andrew Mitchell (ajmitch) said "i'm pretty sure it is a common IDL system" . Perry noted that "with python, you can fudge it and package and unpackage the procedure calls not knowing what they are :)" but "even for non-rpc you still need some kind of .idl file" . Andrew suggested "look at appserver/grpc/GEAS.grpc - or reports/grpc/GRServer.grpc" . Perry said this was just what he had been looking for - "looks like it would work really well with my idea" . He expanded "my "brilliant" idea was that .o files usually contain debugging information that gives you the types of all the symbols - if they are compiled with -g anyway - since you know all the symbols in the file (nm) and you can tell the types (stabs) then you can automatically create a idl file (and, as suggested a .grpc file) and, since most languages can use gnu-rpc of some description (even if it's only corba, or soap or similar) - this means you should be able to write one component in one language (say "C") and another component in another language ("C++"), take the .o files, generate the idl, and then connect 'em all together" .

5. Multi-table Datasources

27�Jun�2002�Archive Link: "[IRC] 28 Jun 2002"

People: John Lenton,�Jason Cater,�Peter Sullivan,�Alejandro Pronotti

Further to Issue�#35, Section�#9� (24�Jun�2002:�Multi-table Datasources) , John Lenton (Chipaca) reported that Alejandro Pronotti (aprono) had managed to implement joins in datasources, with "everything but deletes working, with minimal side-effects on the rest of the codebase - deletes will probably be working before he goes home today" . Jason Cater (jcater) was not sure whether allowing deletes in joined datasources was logical. John admitted that "deletes, updates and inserts don't make much sense unless it's a 1:1 relationship" , adding "if you ask for a combined datasource and it isn't 1:1, you're in for some trouble, i guess" . Jason said "we just probably need a BIG warning in our Developer's Guide - "Only do updates against a join IF YOU KNOW WHAT YOU'RE DOING" :)" . John explained his 1:1 relationships were "for hereditary tables" - "1:1 relations are "*is*" relations" .

John wondered "I wonder what we'll have to hack to be able to add addresses.... and labels probably nothing! :)" . Peter Sullivan (psu) said that was what GNUe Reports had orginially been written for. Jason clarified "well, actually, form letters - but she'll do mailing labels too :)" . Peter tried to claim "from a technical point of view, a mailing label is just a very short form letter ;-)" .

On the join synatax, Jason said he still preferred his second option. John preferred to drop the <condition> tag altogether "and replace it by a 'join' tag, such that it could have a 'type' attribute?" It "could be handled by GConditions" or "maybe something completely different" . Jason was not keen - "I would leave it with conditions - for the time being at least" . John asked about adding "attributes to the conditions" such as "the ones to tell which is the parent of whom, for the deletes" . Jason said this was one of the reasons he had been concerned about allowing deletes in the first place. He felt that joins and conditionals were conceptually different things, "even though the joins will be using conditionals" .

6. NOLA and acclite - free software financials packages

27�Jun�2002�-�2�Jul�2002�Archive Link: "[IRC] 28 Jun 2002"

Topics: Financials (Accounting), Forms, Common

People: Peter Sullivan,�Derek Neighbors,�Andrew Mitchell,�Micah Yoder,�Daniel Baumann,�Jason Cater,�Stuart Quimby

Peter Sullivan noted references to work being done with NOLA, as previously discussed in Issue�#35, Section�#13� (25�Jun�2002:�NOLA as a free alternative to GNUe Financials) , and wondered "Are the GNUe forms for NOLA going to end up being in effect GNUe Financials 0.0.1?" . Derek Neighbors (dneighbo) said not - "we have taken nola - made postgres create scripts for it - retooled directory structure - and gutted some non-free no no's" He was "in process of testing this beast (currently its called acclite) - once it is reasonably tested i will give a tarball to NOLA at which point they can choose to accept and integrate or consider a fork. If they take the patch there will only be NOLA - if the want to fork we will remain as acclite for now" . The intention was "as soon as it seems usable we will start screens for it" although in fact some people had already started. These would either be available "in a non gnue cvs - or check in noguska cvs depending on how things go" . The itch he was scratching was "i have direct go live implementations i need asap that will require fund accounting - so its likely i will look to extend it to add fund accounting pretty quick. I've also got this irressitable itch to use SAP-DB - the more i play, dig read im thinking its by far better than postgres and mysql put together" . Peter asked "are noguska actively developing nola, or is this in effect the main develop stream" ? Derek said "yes they are actively developing - currently they are kind of dead i.e. busy on other projects - but they are looking to hire someone to clean it up and finish it" . He said there was a need for "testers willing to beat on our version so we can send back to NOLA" . Peter volunteered to help out.

Some days later ( , Andrew Mitchell (ajmitch) said he had been "planning to do an accounting system" in PHP and python, but "would prefer not to duplicate lots of work ;)" . Micah Yoder (MicahY) said he was working "on converting web forms in nola to gnue forms." . He said that Derek had done most of the work on cleaning "up the web/php cvs - and I studied some of the code and database schema and made a form" . Derek said "um we need a dcl setup for GNUe NOLA i think - so we can road map what there is to do and such. It still needs testing, testing, testing against our version" .

Micah said "I hate to mess much with the php version - just enough to know what it's doing and be able to convert it to forms" . Derek agreed, but said "we are trying to 'play' nice - which means we need a working php version so i can submit back the changes to them" . He was trying to avoid creating a fork from his side - it was up to Nogasaka to decide whether to accept the changes or not. He personally hoped there would not be a fork, both from the point of view of "publicity (we dont want conotatoin of being leeches)" and of being able to keep the GNUe Forms version and PHP version of NOLA in synch so that people could still use the PHP interface for parts of the system for which GNUe Forms were still being developed. Micah felt that Nogasaku would not want to distribute the GNUe Forms version and their original PHP interface "in same tarball" . Derek said "i imagine what would be logic would be to separate (as already is) the data/schema and ui's - so you could tar each independently if necessary" . Micah agreed, adding "their web system IS pretty nice... some good work was put into the look and style of it...." . Derek said that "lets face it some people are hell bent on having a web system - /me shudders at web accounting, but to each their own" .

Micah did some more testing, and wondered if some of the bugs he was finding were specific to the PostgreSQL version - "in which case maybe I should go ahead and set up mysql for it" . Derek urged him to use the "tar ball from gnue and postgres as thats what needs testing" . The code had been set up in a seperate CVS on the gnuenterprise server, not as part of the normal GNUe CVS on, but snapshots ( were available for people without access to CVS.

Micah went on to try a form for GL Accounts, using the Designer master-detail wizard to include a look-up for Account type. This was just a very small part of a very large overall schema. Daniel Baumann (chillywilly) felt that many of the primary/foreign keys were not set up ideally, probably because "this was dumped from mysql? and then converted?" In PostgreSQL, it was better to "use the REFERENCES word for relations and maintains referntial intergrity for you" . He suggested "the most logical would be to convert the xml db-agnostic format right ;)" as mentioned in Issue�#35, Section�#4� (23�Jun�2002:�Using XML to describe database schemas) , noting that this was "in the TODO :)" .

MMicah downloaded CVS, and noted that "the defines.php file has system-specific databasename/login info" - he would probably just change his database to accept that username and password rather than change the code. However, it still had flags set incorrectly - "that is what requires the non-free fonts. - I had to set them to zero to make it work" . Jason Cater (jcater) said "that can be your first commit :)" , noting "what you see in that cvs right now is the result of 4 hrs of my time" , and welcomed any input people could give. Daniel downloaded the code, and noted "PHP is fugly next to python" . Jason said "I hope ppl only spend enough time in the PHP version to figure out how to do a Forms version" . Daniel was impressed that pie charts were included.

After midnight ( , Micah noted that the PostgreSQL schema did not have default values set for any field - "I guess the script missed them. :(" . Derek said this is why he wanted to use the XML schema extraction tool mentioned in the day before, which should include everything except indexes. Micah noted that there were other problems with adding new account types. Derek said he had seen "mail from stbain that said this was beta (even their stuff) - its quite possible no one has tried to add additional types yet" . He felt he "should setup NOLA + MySQL out of box then setup Acclite + Postgres" - "so can do side by side comparisons when there are issues" .

Micah did some more testing, having to switch from Mozilla to Konqueror due to some odd behaviour with the "back" button. Peter felt that "in db-based web apps back button can be misleading as people think it = undo - but you've still committed your changes" . Micah agreed, but said "I know what I'm doing and want to use it. :)" .

Later, Stuart Quimby (ToyMan) asked "you guys are gutting nola *really* thouroughly, yes?" . Jason said "not really gutting - as much as making it work with postgresql instead of mysql" . The "first step is to get schema into postgres so it works with current php logic - then when we have the schema down we create replacement GFDs for the PHPs - that's the plan at least" .

Later, Micah reported problems "trying to get it to add vendors" . After some investigation, Jason suggested that the update was failing because the names were stored as char fields rather than varchar, and hence the trailing spaces were breaking the match. Derek said he "forgets people even still use CHAR for anything - definitely we should be using varchar for postgres" . Micah said he would also "go through the schema and add 'default now()' to the timestamp types that don't have them - as that was an issue in a couple places" .

The next day ( , Andrew noted a "security hole for NOLA on bugtraq" - "says that users can upload what they want thru upload pages - NOLA does no checks (they can upload php files, etc)" . Jason said that this was not as serious as it sounded, "as anyone who runs a publicly accessible NOLA install (accounting, mind you) just aint right" . Applications like this would normally be very much inside the firewall. For internal users, "/me isn't too worried about my accountants h4x0ring their accounting package" .

Later, Micah asked "is there a place on the web to find cvs commits for acclite?" He did not want to subscribe to "the commit maillist" as he already got "too much crap in the mail. :)" Peter said that there probably was not, as the acclite CVS was on the gnuenterprise server, not

7. Schema extraction wizard in GNUe Designer

27�Jun�2002�Archive Link: "[IRC] 28 Jun 2002"

Topics: Designer, Common

People: Jason Cater

Further to Issue�#35, Section�#4� (23�Jun�2002:�Using XML to describe database schemas) , Jason Cater (jcater) announced he had added schema extraction to GNUe Designer "via a Wizard - reads table definitions from the database - and creates a GNUe Schema Definition" . This worked with any database driver in GNUe Common (GNU-RPC) "that supports introspection" . It "currently only grabs tables/fields/default values/nullable settings - does not grab sequences/triggers/primary keys/foreign keys" as of time of writing - "that's really the biggest deficiency of it right now i.e., it doesn't extract integrity constraints - but, heh, this is only a few hours of work so far ;)"

8. PyXML no longer a dependancy for GNUe

28�Jun�2002�-�29�Jun�2002�Archive Link: "[IRC] 29 Jun 2002"

Topics: Application Server

People: James Thompson,�Andrew Mitchell,�Jason Cater

It was reported that GNUe's CVS install was still giving an error message saying that PyXML was not installed. James Thompson (jamest) confirmed it was no longer required, as previously discussed in Issue�#21, Section�#18� (20�Mar�2002:�PyXML dependancy removed) - "we must have missed removing that test as the xml included in python 2.x is now supported. However some distros (like debian) split the xml off for some odd reason" . He would check this later, and remove the unneeded checks if they were still there.

The next day ( , Andrew Mitchell (ajmitch) reported that Application Server was asking for "the PyXML package" . Jason wondered why - he had "worked hard to get rid of that dependency" . Andrew said "dunno, but it chokes without it :)" .

9. SAP-DB community

29�Jun�2002�Archive Link: "[IRC] 30 Jun 2002"

Topics: Why GNUe?, Common

People: Derek Neighbors,�Jason Cater

Derek Neighbors (dneighbo) said he was "on mission to check out sap ( much closer as well - as i found out they have OLAP functions too (or so it seems)" . He noted that SAP-DB had an 'Oracle mode' "that lets you use ORACLE SQL QUERY syntax out of box" . Later, Jason Cater (jcater) said that query syntax was not the main problem - "my big hangup is 1) their triggers are differnt than oracle's" .

Derek said he had been really impressed with the sense of community that was developing around the SAP-DB now that it was available under the GPL (GNU Public License) - "someone was trying to start a user group - and i said they really needed a good sap portal with tutorials, articles, howto's, samples etc - the SAP guy responded and gave url to a wiki we could use (me doenst like wiki, so sue me) - and said he wished sap would do something better, but it was resources issues and policy wouldnt let them give access to anyone outside sap to making pages - but then within 12 hours someone" had "created a portal (postnuke i think) from the thread" at Jason said it was "not bad for something that was put up today" . This inspired him to ask whether SAP had a sample database in the same way that Oracle did. Derek said he was "pretty darn sure they do (a large one iirc)" . Jason suggested writing some sample GNUe Forms to run against the SAP-DB sample database.

10. Read-only fields in Forms

31�May�2002�Archive Link: "[IRC] 01 Jul 2002"

Topics: Forms

People: Arturas Kriukovas,�Derek Neighbors,�James Thompson

Arturas Kriukovas (Arturas) reported "finally have fixed readonly entry input bug!!!" Later, Derek Neighbors (dneighbo) noted "imho readonly fields should work as follows - when you are in 'data' mode (normal operations) you should not be able to enter data into them, and preferably they would not be part of the tab order - if you are in 'prepare query' mode they would allow entry (for searching purposes) and preferably would become part of the tab order" . James Thompson (jamest) confirmed "that's how they worked last I knew" . Derek said "there was bug filed - a readonly widget behaves oddly... i.e. if you go to a readonly field in 'normal' mode it woudl let you type but what you typed would randomly appear in other widgets" . Arturas confirmed he had updated CVS to fix this, and it still worked the way Derek said - "you can enter data in readonly entry in query mode - you can not enter data in readonly entry in data mode (message box pops up and says "You can't do this blah-blah") - in query mode readonly entries are part of tab order" .

11. Python encoding in GNUe

31�May�2002�Archive Link: "[IRC] 01 Jul 2002"

Topics: Forms

People: Dmitry Sorokin,�James Thompson,�Andrew Mitchell,�ra3vat

Dmitry Sorokin (ra3vat) noted that the Windows "binary forms snapshot is build with defaultencoding set to ascii is it possible to change it somehow after installation?" James Thompson (jamest) said that there was "a gnue.conf option now" for database encoding. Dmitry said he was talking about the python getdefaultencoding(), which was not checked again after "python startup process" . He found the code in the module that was setting this, and suggested changing the source code, as he could not find "a way to chande defaultencoding later" . James agreed - "we'd have to just add that to our build instructions for win32 - and I don't see how it'd break anything" .

Dmitry asked "is it hard to rebuild? i'd like to test it" . Andrew Mitchell (ajmitch) said that the windows installer used mcmillan ( and inno ( , as previously discussed in Issue�#10, Section�#5� (30�Dec�2001:�Using McMillan to distribute python executables) . James confirmed that "there is a text file giving step by step instructions on how to do it" - if you wanted to compile ODBC support in as well, you needed the ODBC drivers too - "it's from activestate IIRC" .

James wondered "why python hard coded ascii vs the defaultlocale" . Dmitry said that "with regular installation it is easily workarounded with sitecustomize - hack on would work also" .

12. Deleting records from within triggers

31�May�2002�Archive Link: "[IRC] 01 Jul 2002"

Topics: Forms

People: James Thompson,�Derek Neighbors,�Jason Cater

James Thompson (jamest) asked "i need to mark records for delete from a trigger - what would people feel is the best way to make this work?" After some discussion, Derek Neighbors (dneighbo) felt "/me doesnt like the logic of 'marking' in code as transactions (commit/rollback) etc shouldnt distinguish" - a rollback should roll back everything since the last commit, whether it was an insert, update or delete - it "shoudl not be something developer worries about in the detail" . Jason Cater (jcater) asked "why are you playing with transactions?" Derek said "well to me 'commit' implies transaction - even if its at form level and not db level" . After some more discussion, Jason said "so, in a nutshell, you are saying use block.delete() instead of block.markForDelete() ? which I would tend to agree with" .

13. Modal windows in Forms

31�May�2002�Archive Link: "[IRC] 01 Jul 2002"

Topics: Forms

People: Marcos Dione,�Derek Neighbors,�James Thompson,�Jason Cater

Marcos Dione (StyXman) asked "do you think that make modal windows from xml files woud be: a) a good idea. b) possible?" Derek Neighbors (dneighbo) said he "would like to maybe see them like 'wizards' in reporter i.e. easy for users to make new ones - but not really just xml files" . However, it depended "how complex of a modal box you are looking for" - "more often than not they will be done via triggers" . Marcos said "we thought that xml files would be great, because we already know how to handle those" . Derek "generally doesnt see modal boxes as being data driven nor having widgets really" . Marcos said "I guess what we'll need more are input modal windows, ones that just makes a question to the user and that value is teturned to the main form." Derek gave some examples - "1. is a search dialog - 2. is a picker dialog" . Marcos added "3. data input, like passwd ones" . This would be for situations after the normal login to provide extra security for "a superviso auth a non regular trabnsaction" such as over-riding normal credit limit checking.

James Thompson (jamest) said "i wouldn't want to have to define windows - but I think i'd be ok with a trigger creating a dialog box that prompts the user for info - i just don't think we'd want to start laying out lots of window defs in the xml" . Derek feared "it would be UGLY to have to create with triggers" and emphasised "im not saying it has to be xml layout" . James noted that "the search/picker is planned" - it would be standard Forms functionality that could be called "in the sql statement - and it creates the picker" .

Derek said tha "if lots of people start wanting different modals - want to 'enhance' modify current modal" then having these as templated definition files (whether XML or not) "makes more sense as it keeps people out of core code" . However, he was not sure how common this would be. Modals like a picker were fairly simple, "but search, you will get 100 different ways people want that implemented" . Marcos felt that "if you give them a standar way to implement search and picker, I don't think they'll make another one. unless the ones you give are not general enought."

James suggested that "searcher/picker would be coded in backend" and called by tags in the XML for the GNUe Forms Definition. Jason agreed - "the modal would just be another <page>, but with a type="popup"" . This was added to the TODO file for either Jason or Marcos to implement when they had time.

14. Preferred python driver for PostgreSQL

1�Jun�2002�Archive Link: "[IRC] 02 Jul 2002"

Topics: Common

People: Chad Walstrom,�Jason Cater

Chad Walstrom (^chewie) asked "is the "suggested" python api for postgresql? I've seen a few python projects using it now, but I've seen very little support for it on, say Debian" . Jason Cater (jcater) noted that "the PyGreSQL ppl created which is a pretty impressive library" but had had the DB-SIG compatible features added later, as opposed to "psycopg or pypgsql" which had been "designed fomr the ground up to work as a DB-SIG module" . He emphasised that he was not "dissing the PyGreSQL ppl in any way - as their original module was/is pretty feature-complete - just non-standard" .

15. New web site for GNU Enterprise

2�Jun�2002�Archive Link: "[IRC] 03 Jul 2002"

People: Derek Neighbors,�Peter Sullivan

Derek Neighbors (dneighbo) said he had looked at the new draft website for GNU Enterprise - "it looks REALLY complete" but he "had a few questions though about maintainability" . Peter Sullivan (psu) said "like everything else, it will need some docs ;-)" - he was "working on a README" . He added "the main maintainability (hmm) issue at moment is news items" , which needed some kind of automatic submit/approve mechanism. Derek noted "the status items and some other things are on each tool/package page as well as on the page the item belongs (like status) - is that being typed twice" ? Peter said "at the moment, yes - I need to make that more "clever" - have a status("tool") object that returns relevant text" so that it only needed updating in one place.

16. Checkboxes bug on Forms

2�Jun�2002�Archive Link: "[IRC] 03 Jul 2002"

Topics: Forms

People: James Thompson,�Arturas Kriukovas,�Bajusz Tam�s

James Thompson (jamest) said he had "unpatched that patch" which Bajusz Tam�s (btami) had submitted and Arturas Kriukovas (Arturas) had applied, as it had broken CVS. Arturas said the patch was intended to fix "some bug on win32, some aspect (?) of checkboxes" . Later, Bajusz apologised for the typo in the code. James said he had tried to fix it, "but I didn't know what you were fixing so I didn't know exactly what the code was to do" . Bajusz said the original bug had been that "self.display type is int on checkboxes" . James agreed that "checkboxes are evil :)" and said he needed "to gut them sometime soon" .

17. Calling methods on the Application Server

2�Jun�2002�Archive Link: "[IRC] 03 Jul 2002"

Topics: Common, Application Server, Forms

People: Jan Ischebeck,�James Thompson

Jan Ischebeck said he wanted to "add a special 'call method' function to the appserver dbbackend" that could be called in a trigger by a GNUe client such as Forms or Reports. It should probably work "like the getSequence etc. in the postgres" driver. James Thompson (jamest) said that there was already code "for Trigger Exposure - you'll see a dictionary that contains exposed function names for the triggernamespace. self._triggerFunctions has an entry for the sequence functoin that calls a functoin that then uses the triggerExtension" . Jan said this meant that "if I add functions to triggerExtensions they won't be added to the triggernamespace automaticaly." . James claimed that "triggerExtensions are old school :)" but admitted they "still have a place" . He pointed to some similar code he had written. Jan said he "could add an entry to self._triggerFunctions ... and even make it global, but what do you think is the best place and name for a function that calls the call function of one instance" ? James said it should search "the loaded drivers triggerExtensions for a call method entry and if it existed it could call it - and if it didn't exist then it could throw a not supported by backend error (or something)" .

18. Capitalisation for wizards

2�Jun�2002�Archive Link: "[IRC] 03 Jul 2002"

Topics: Designer

People: Arturas Kriukovas

Arturas Kriukovas (Arturas) pointed out a comment in the GNUe source code which said that "It is VERY important that in any reference to a "<insert type here> wizard", the word "wizard" must be in lower case, as many "Xyz Wizards" (proper name) have been trademarked. (No, I'm not kidding :)" , saying he had not been aware of this.

19. Referencing the current record in Forms

2�Jun�2002�Archive Link: "[IRC] 03 Jul 2002"

Topics: Common, Forms

People: Jan Ischebeck,�James Thompson

Jan Ischebeck (siesel) asked "how can I get the current record?" . James Thompson (jamest) said "the _form should always contain a _currentRecord - but I'm not sure it's correct" . Jan suggested using triggers, but James was not keen "as then it'd be Forms specific - if you can keep it in datasource then it'll work for all gnue apps" . Jan reported "seems like I can the currentRecord by dts.getRecord(dts.getRecordNumber())" . James said he had tried something similar, but had not been able to get it to work consistantly, "so added the posting record var" as a quick fix. Jan confirmed his solution worked for normal triggers. James suggested testing with pre-commit and pre-insert triggers as well, warning "during a commit the datasources don't keep in sync with the UI" .

Sharon And Joy

Kernel Traffic is grateful to be developed on a computer donated by Professor Greg Benson and Professor Allan Cruse in the Department of Computer Science at the University of San Francisco. This is the same department that invented FlashMob Computing. Kernel Traffic is hosted by the generous folks at All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License version 2.0.