"i always appear in kc as some sort of insane topic drifter knowitall" - "if the shoe fits..."
Table Of Contents
Introduction
This Cousin covers the three main mailing lists for the GNU Enterprise (http://www.gnuenterprise.org) project, plus the the #gnuenterprise IRC channel. Above headquote specially chosen for this, the first anniversary issue of Kernel Cousins GNUe. After a year in which the GNUe Tools have taken most of the attention, the near future may be more focused on the GNUe Packages, as foreshadowed below.
1. Volunteers for General Ledger
23�Oct�2002�-�28�Oct�2002 (6 posts) Archive Link: "If you are volunteering we can certainly put you to work."
Topics: Financials (Accounting)
People: Lawrence,�Louis Charbonneau,�James Thompson,�Derek Neighbors,�Todd Boyle
Lawrence asked "Where does gnue need the most help? I am looking for a project to get invovled in" and would have some spare time soon.
Louis Charbonneau said "A year ago , I made a proposal for the Financial part of GNUE. But I had to quit for personnal reasons. Now , I am back to this project. I could team up with Lawrence, and start working on GL module for example. What do you think ?" James Thompson welcomed Louis back, and said "I'd presonally like to see something we could implement in both a traditional client/server setup (like us old timers use :) and something that we could also implement in appserver (the replacement for GEAS)." . Derek Neighbors said "I think this is great news. Louis I loved your original proposal and writing style. It will be good to see you hard at work. :) We are ramping up some major stuff with regards to 'packages'. I would like to coordinate these efforts. In relation to financials. We have a based built from NOLA web accounting called AccLite. It would be great to have a business analyst/accountant give it a good testing and evaluate if its a good base to start from or if we should start from scratch" - "it looked like its structures were somewhat decent and that we could probably build rather quickly forms screens to match them and have a GNUe Application from that base. I have oodles of manufacturing and CRM type stuff Im working on and Jason has several applications that he is working on. We would like to get these into the populous and coordinate efforts on building." He suggested discussing further in IRC, if possible.
Todd Boyle also offered to help - "requirements must be observed at several levels. Goals and requirements at a business or "enterprise" level, and at technical and infrastructure levels. I would like to contribute to definitions of ledgers and their various methods, elements of the content such as charts of accounts, etc. XBRL taxonomy for GLs has not done a bad job of enumerating most of these things now, let's help them out with an implementation. An implementation that understands and embraces not just "mapping the content", but interoperability of GLs. Let's build a GL that is compatible and interoperable with *every* reasonable GL in the world, commercial or otherwise."
2. GNUe support for Oracle 9i
24�Oct�2002 (2 posts) Archive Link: "[Gnue-dev] Configuration of GNUe + Oracle 9i"
Topics: Common
People: Jason Cater,�Derek Neighbors
It was asked how to connect GNUe to Oracle 9i. Jason Cater said "You will need to install DCOracle2 (http://www.zope.org/Members/matt/dco2/DCOracle2-1.2.tgz) (a Python Oracle driver) on your" client "machine in order to access Oracle." "Untar this file and run "python setup.py". Be sure your $ORACLE_HOME variable is properly set." This assumed that the (proprietary) "Oracle OCI8 client libraries" were already on the client.
On IRC (http://www.gnuenterprise.org/irc-logs/gnue-public.log.25Oct2002) , Derek Neighbors (derek) said "my question is people looking for ERP that love oracle, why not looking at compiere ;)" as previously discussed in Issue�#32, Section�#1� (27�May�2002:�Freedom issues with using Compiere as an alternative to GNUe) - personally, "even if i had oracle i would want gnue :) - but most people dont care about doing it right - they just want instant gratification" . He explained that Compiere was "an 'open source' ERP built on oracle 9i and java - it isnt a framework like gnue but its packages are far more complete" . However, "long term you are better off with gnue (and we do work with oracle)"
Jason (jcater) cut and pasted his connections.conf file to show how an Oracle connection should be defined for GNUe - Oracle used the concept of a "service" rather than a host name and database name. The service then had to be defined in Oracle's tnsnames.ora file on each client.
3. Bug-squishing in Forms
23�Oct�2002�Archive Link: "[IRC] 24 Oct 2002"
People: Bajusz Tam�s,�James Thompson,�Ariel Cal�
Bajusz Tam�s (btami) reported "one bug remained in event rework - CreateWidget is missing from GFEvent.py" now that the event handling code had been moved to GNUe Common. (This had been done so that other tools, notably Reports, could re-use it.) James Thompson (jamest) asked for someone "with a working copy of cvs gnue-forms" to "test the forms/sample/intro/intro.gfd form - enter a year, click the button and tell me if the fortune only blinks on the screen for a second or if it stays up - /me is trying to figure out if he broke something" . Later, Ariel Cal� (ariel_) reported "fortune stays well and alive" . This was with "today download from cvs but may be stup-cvs yesterday" - but still worked even when he "rebuilded" .
4. Version of 0.4.0 GNUe Tools for Microsoft Windows without debug window
23�Oct�2002�-�24�Oct�2002�Archive Link: "[IRC] 24 Oct 2002"
People: Bajusz Tam�s,�Peter Sullivan,�Keith Jagrs,�Jason Cater,�Keith
Further to Issue�#52, Section�#6� (18�Oct�2002:�Fixes to 0.4.0 releases setup.exe on Microsoft Windows) , Bajusz Tam�s (btami) asked "psu: have you got my e-mail?" Peter Sullivan (psu) said "not yet. If it's what I think it is, it could easily take an hour to download" on dial-up. He said "well, KeithJagrs will be pleased ;-)" He explained "I managed to "persuade" btami to do us a Win .exe without the debug window as well" . Bajusz explained this was just a case of setting "console=0 in *.conf files" for McMillan, the package GNUe used to convert python into stand-alone Microsoft Windows executables. Keith Jagrs (KeithJagrs) said "Maybe its better to have the debugging console. Specially in this early stages of gnue tools. so you can point out the bugs" . Peter said "I think it depends. If, like you, you want to put it in front of end-users" , debug probably needed to be off. Keith said "yes. but until now I've been just playing with forms or designer." .
The next day (http://www.gnuenterprise.org/irc-logs/gnue-public.log.25Oct2002) , Peter Sullivan (psu) announced "for the record, both the downloads/current and downloads/releases dirs now have both a debug-on.exe and debug-off.exe for Win 32 of the Tools 0.4.0a releases - thanks to" Bajusz Tam�s (btami) "for all the hard work" . Jason Cater (jcater) queried the "0.4.0a" designation. Peter explained "well, release 0.4.0, package a - i.e. still based on the 0.4.0 release tarball - but with "no wonder jcater hated mcmillan, but we finally got it working" packaging" . Jason said "it's a love/hate relationship" ...
5. Application Server documentation on web site
23�Oct�2002�-�25�Oct�2002�Archive Link: "[IRC] 24 Oct 2002"
Topics: Application Server
People: Reinhard M�ller,�Peter Sullivan,�Derek Neighbors
Reinhard M�ller (reinhard) noted that the web page for Application Server "seems to refer to old geas" . He suggested "a) rename the page to appserver.php and b) take the text from appserver/README" . Peter Sullivan confirmed "the filename is wrong - same for /tools/gnurpc.php -> common.php" but he had not realised that the text was wrong too. Reinhard also suggested that the whitepaper and API documentation for the new Application Server should be added to the docs page. "IMHO it would make sense if" at least "the html version was online" . HTML, Postscript and PDF versions could all be created using make.
Also, Reinhard "would like to see the "GNU Enterprise: Module Design Proposal" marked as "obsoleted"" . Derek Neighbors (revDeke) thought "i think its still pertinent - it needs updating but i plan to revive it very soon as i plan on pumping out a lot of modules - and i want something as my template... though i might replace it with an open office version" .
Reinhard said that "if i can help you with the text (for appserver.php) or the docs please tell me. not sure if there is a possibility that i could update the docs on the web page myself as often as i change them" . Peter said "we should have the web site in CVS real soon now" .
Two days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.26Oct2002) , Peter said "I have now uploaded the whitepaper and api docs to the website. Had a bit of trouble, as I couldn't find texi2pdf in my woody distro - so in the end I did the pdf and ps versions from the html using htmldoc instead. They don't look too bad that way" . Later, Reinhard noted that texi2pdf was in the "tetex-bin" package.
6. GNUe Roadmaps and general project organisation
23�Oct�2002�Archive Link: "[IRC] 24 Oct 2002"
People: Keith Jagrs,�Peter Sullivan,�Derek Neighbors,�Nick Rusnov,�Daniel Baumann,�James Thompson,�Keith
Keith Jagrs (KeithJagrs) said "I'd like to help in the definition of roadmap. I think it's essential" . Peter Sullivan (psu) agreed, but said that the "Biggest problem will be that we can't realistically put dates on" but "we have cross-dependancies." For instance, "we didn't really want to start on apps until we have a prototype app server" . Although this particular issue was resolved, further ones would occour as features in the packages became dependant on features in the tools.
Keith suggested "Maybe it is better to put efforts on tools first, then in packages" . Peter said "that's sort of what we've done, I guess - the Tools are not pefect, but I think they belie their 0.x.0 or 0.0.x version numbers in terms of kick-ass functionality" . Keith said "so in terms of commercial apps, what would be some more accurate version numbers for the current releases?" Derek Neighbors (revDeke) felt "if windows is on 2000 forms is at least at version 10.0 - compared to other opensource (sic) stuff we are probably somewhere between 1.0 and 2.0 - it all depends" . However, "apache, gtk and the likes were good about version numbering and so i think 0.4.0 is realistic if you use them as a baseline - if you use other projects claiming to be version 8.0 and such we are probably similar" .
Derek said "i have been screaming roadmap for a bit - its not that simple though" . "we have multiple things going on - so its hard to compile a road map - as things are interdependent - so you need to do all roadmaps in a cohesive manner" . Keith felt that a "roadmap should be done on concensus. not one persons effort" . Derek agreed, but "thats huge task :) with lots of players - i suspect someone will have to compile the TODOs of the tools and such - and make a draft roadmap and solicit feedback - and modify the feedback into the roadmaps, rinse and repeat a few times - and there is something go by. The other issue is getting a road map might not be too bad for the 'next' release but a good roadmap will include 2 to 3 releases out" .
Keith, reprising Issue�#52, Section�#9� (19�Oct�2002:�WikiWikiWeb for GNUe) , asked "what about setting up a Wiki?" Derek said "i think wiki is good for user driven content that needs minimal controls - but official documentation, guides, roadmaps and such - that need tighter review and control are not good for wiki IMHO" . Nick Rusnov (nickr) noted "most wikis these days allow you to have at least access control." Peter said "we already have 10 channels of communication" such as mailing lists and IRC - "/me would support a wiki if it replaced some of those - rather than just became number 11" .
Keith said "I favor a webforum for instance, instead of a mailing list. (but thats is just my opinion)" . Peter said "we tried that for about a year - I was responsible for killing it as no-one was using it" . Daniel Baumann (chillywilly) said that was "not exatly the same thing - PHPNuke is like a news site - PHPBB or something similar would be a forum" . Peter said "the real focus of this project for good or ill is IRC - mostly for good, IMHO" . Daniel agreed - "we have IRC so this is our support system ;)" . Keith felt that "as the user base or contributors grow, a forum is better suited than a mailing list" . Derek said "webforums suck for developers - i have no problem people making what works best for them - but dont expect developers (those most capable of helping) to spread accross 15 different mediums" . Nick Rusnov (nickr) felt that "mailing lists are great because they are easy to ignore - it takes no effort" . Keith said "not for developers then. but for discussions on features an such. and roadmaps" . Web forums were "easier to browse than a mailing lists archive" . James Thompson (jamest) said "we did forums, long long ago at peoples insistance (is that a word? :) - anyway, we set them up and they got used exactly ZERO times - kinda ticked me off to be honest :)" Peter suggested "the best way is to hard-wire our brains together in a neural net - then we'd all finally understand chillywilly" . James said "no way I want to know what goes on in chillywilly's brain man - i barely can keep track of the voices in my head let alone his" .
7. Data modelling and schema tools for GNUe
23�Oct�2002�-�27�Oct�2002�Archive Link: "[IRC] 24 Oct 2002"
Topics: Designer
People: Nicholas Lee,�Daniel Baumann,�Charles Rouzer,�Charles Rouszer,�Jason Cater,�Jan Ischebeck
Further to Issue�#52, Section�#5� (17�Oct�2002:�Free Entity Relationship Diagram tools) , Nicholas Lee (esands) asked "Anyone suggest a good data modeller? There doesn't seem to much free stuff" . Daniel Baumann (chillywilly) suggested "dia ;)" . Nicholas said this "doesn't generate sql though. I just managed to get pgdesigner (http://www.hardgeus.com/projects/pgdesigner/) compiled. Simple but looks useful" He had also looked at "http://www.alzabo.org/ (perl nightmare to install, needs mason)" . Charles Rouzer (Mr_You) wondered "if you could build that into GNUe. maybe one day ;-) maybe GNUe Designer could have "New->Form" and "New->Schema" or somethin." Nicholas said "I guess someone could - pgdesigner is in c++ - Problem is with SQL is that different DBs have different requirements" . Charles said "well, I guess it would only be useful for designer. unless you wanted to be able to create tables from a Form" . Nicholas was concerned about the security implications of this. Charles said "well.. to create/edit Forms it requires you have some sort of ability to build a database table.. as things get complicated a schema designer is a good addition to Form building. it would fit right in with Designer. right now I think it reads schemas" - "actually I guess it just reads individual tables. I believe the wizard will query your datasource (db table) and help build a form" . Nicholas said "I guess in that sense, doesn't help you put together then schema though" . Charles agreed - "the schema would mean multiple tables.. so now we have another wish list item ;-)" . Daniel Baumann (chillywilly) said "there's multi-datasource in the works already" - he had seen some commits to CVS about this.
Some days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.28Oct2002) , Daniel Baumann (chillywilly) asked "does designer do db schemas yet?" Jason Cater (jcater) said "not well - it still has a ways to go" . Nicholas Lee (esands) said "supposely with some xsl and dia you can. pgdesigner can do the job as well" . Daniel said "we have an xml schema format and" Jan Ischebeck (siesel) "was working on integrating things - it is borrowed from dcl I believe" .
8. Problems with Forms splashscreen
24�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
Topics: Forms
People: Lawrence,�Jason Cater
Lawrence (lawrence) reported
"still geting
a Segmentation fault with gnue-forms - Splash screen comes up; then
font chooser; then segv - gnue-designer is fine"
.
Jason Cater (jacter) asked
"can you try
running with the -s option, and seeing if it still segfaults?"
.
Lawrence confirmed
"looks like -s option did the
trick!"
Jason said
"you probably
want to modify your gnue.conf file with:
[forms]
disableSplash = 1
- then you won't have to use the -s option all the time"
.
He added
"we are having some issues
with timeouts on our splash screen - it rears it's ugly head on
slower machines. I guess we either need to find a solution, or
disable the splashscreen by default :("
He explained
"the problem is, if the splashscreen
tries to disappear before the main screen is completely drawn
it segfaults. We can increase the timer more - but that's
avoiding the problem, not fixing it :)"
Lawrence said
"slow machine? 1.4 athalon with
a gig of memory not fast enough?"
Jason said
"wow - this is the first time we've seen
this problem on any thing faster than a 266Mhz"
.
9. Python site customisation for encoding
24�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
Topics: Forms
People: Bajusz Tam�s,�Arturas Kriukovas
Bajusz Tam�s (btami) pasted some code he had added to his "sitecustomize.py" file to avoid having "to set different encodings" . Arturas Kriukovas (Arturas) asked if this meant "offering final user to modify his/her sitecustomize.py to enable GNUe work with his/her encoding?" Bajusz said "no and yes :) - i imagine that setup process puts sitecustomize.py into some right place" normally, rather than the user setting it every time. Arturas noted that "one linux guru noticed, that forcing users to modify sutecustomize.py is no good for several reasons - the most simpliest case would be some other software that needs current sitecustomize and it can't work with other sitecustomize and GNUe, that can't work with current sitecustomize - it's good for us as developers - it's good for some users - but we can't enforce this - and we should be able to support encodings as well as possible with current sitecustomize.py" . Bajusz said "maybe we can offer this if there isn't any sitecustomize.py" existing already. Arturas said "[i _think_ (i don't know this for sure) that installing python installs sitecustomize.py]" automatically. Bajusz said "in win32 and python 2.1 2.2 the file doesn't exist per default - and i think forcing enduser to make this file then write somethong into it" was not very user-friendly - "plus he have to know what is your encoding name" . Arturas said "the idea is pretty good" but at the moment, his priority was getting a "working setup-cvs.py for windows :) :) :)" , as discussed in BROKEN KCREF and previous threads.
10. GNUe Debian packages won't install correctly over a non-package install
24�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
People: Nick Rusnov,�Arturas Kriukovas
It was reported that the GNUe Debian packages did not seem to work - the error message (a python Tracebabck) was pasted. Nick Rusnov (nickr) said "that indicates serious package breakage." Arturas Kriukovas (Arturas) asked "you have all the dependencies, don't you?" but Nick thought it was something more serious - "I don't think that error would occur if there were missing dependencies" . Later, the solution was uncovered. A previous version of GNUe, installed directly rather than using the Debian packages, was already on the system - removing this first solved the problem.
11. Updating drop-down lists and changeable runtime parameters
24�Oct�2002�-�28�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
People: Derek Neighbors,�Jason Cater,�Jan Ischebeck
Derek Neighbors (revDeke) asked about adding support for "the realtime updating of dropdown boxes" , as discussed previously in Issue�#32, Section�#27� (3�Jun�2002:�Look-up combo boxes in Forms) . He explained "what i have is a case of product categories that can have parents - the parents are in combo boxes - so if i add a new one - then add a second and want what i added as the first to be a parent, i have to shut the form down and restart it to get it to show in the dropdown" . He would not object if the solution involved doing "it via triggers" . Jason Cater (jcater) said "I'm adding a resetForeignKey, but do note that you are programming for specific UI features if using a combo that way" .
Later, Jan Ischebeck (siesel) said "I've tried to build a testcase for that resetForeignKey() statement, but there are some issues, which need to change a bit of code in common" - "its the parameter stuff which made me changing stuff. i've made a form with two dropdowns. The user can choose a state in the first, and should choose a city in the next and will get displayed the zipcode in an style=label entry. The second dropdown is using a parameter in a condition to display just the cities from the state chosen in the first dropdown. So I added a setParameter function. But that didn't work because of caching parameters values in the cparam class and because of the dbsig dbdriver replacing the staticCondidition by a preparsed string :( - which is also the reason of passing a parameter in the command line is not working at the moment." Jason said "hmm - I had intended on runtime parameters to be read only - /me is trying to think through your example" . He said "I guess they don't have to be - I suppose we can add a setParameter and get rid of the caching code" . Jan said "other possibilities would be to have a 'populateForeignKey' or a 'add/change staticConditions'" . Jason doubted that "the caching code speeded things up much anyway" .
Jan pointed to the problem line of code he wanted to change - "This will change the GObj tree." Jason wondered "should we NOT modify the GObj tree in _dbsig - but still create the static condition - but add an invalidateCachedConditions() method" . He explained "it recursively changes stuff until you are left with a single string - so that it only has to do that once" - "in the end, there's only one node" . This meant that "there's not a clean solution to" adding support for "changable runtime parameters" .
Jan said "the other possibility would be to change the staticConditions using a trigger." Jason said "so you are thinking we could add a setCondition(<GCondition tree>) method to GDataSource? or something like that" ? Jan agreed - "its just a bit OVERKILL for just changing a parameter - I mean, I would like to have both possibilities." Jason agreed, but "what you are describing is not" the "typical use of a parameter - although I'm not opposed to having a setParameter - as long as we don't add much more overhead to datasources" . He gave a specific example - "I have one report using GNUe Reports right now that's a master/detail report - there's approx 1,800 records in the master datasource - so that's 1,800 times the child datasource is run" - and "I have much more complex reports to do" .
Jan said "my actual patch to have parameter working is just to NOT cache the static part of the parameters - If reports is getting too slow we still can add caching, and probably in a better way :)" Jason said "so, the plans are to add a DataSource.setCondition() and a Form.setParameter() and remove parameter caching? We can see how that works. I may be worried about nothing :) - but you do understand my worry about reports, right?" Jan said "I fully understand. And I finally know why you add that caching line to parameters" .
Some days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.29Oct2002) , Jason said he was "almost ready for a 0.4.1 release of stuff" . Derek asked "did we come up with a way to update lu comboboxes? we had discussed using triggers or the likes" . Jason said "it's not fully tested - but we have an Entry.resetForeignKey()" . Derek said he had a form he wanted to test it with - "and if it works for me you can probably put in 0.4.1 ;)" . Jason asked "how does this form work? as that determines where you'd put it" . Derek said the table had "id, description, parent" but some ids were also parents - "basically the category table is a foreign key to its self" . Jason said "we can do this - BUT any new "childs" will need to be committed to the database before we can get the dropdown to update" . Derek agreed - "i was going to put the resetFK in the postcommit" trigger.
He tried it, and reported "didnt work, BUT didnt get a bunch of errors :)" Looking at the debug messages, Jason said "the trigger is working - wx is just not refreshing properly" because of "the way we use it" . However, "this is good though - you now have the form the way it needs to be - it's just we have to get forms" client " p to par w/the" GNUe Forms Definition (.gfd). Derek asked "how hard would it be to fix the resetkey to update widget" ? Jason said he could "probably do tomorrow - if I have a sample form to play with" .
12. Sample/demo database for GNUe
24�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
People: Lawrence,�Derek Neighbors,�Jason Cater
Lawrence (lawrence) asked "is there a gnue reference database used for demos?" Derek Neighbors (revDeke) said "most of the samples have sql create scripts to create teh database to run the demos - currently they are for postgres BUT i plan (or someone else could) convert them to use GNUe Schema Definition - in which case all dbs woudl be supported :)" He added "im thinking maybe we make gear files and navigator files - and one of the launchers on the navigator runs the actual db create scripts?"
Later, Lawrence asked "Does any know of a good reference database that would work well with gnue? MS uses Northwind" . Jason Cater (jcater) said "we need one - I was thinking about that when writing the developer's guidfe - it would make samples much easier" . Lawrence asked "does SAP-DB come with one?" . Derek replied "yes i believe it does" . Lawrence said the scott tiger sample database used by Oracle "is not that complete last time i looked" . Jason said that "scott/tiger is a good start - that's one thing I liked about Oracle's books - every one of their manuals (and boy did they have a LOT of manuals) all used examples based on the scott/tiger db - I'd like to see us do something similar" . Lawrence said "I'll work on one - does any one have suggestions" ? Jason said "well, if possible, it shouldn't be country-specific - like "zip codes" etc - (that's the example I used in the developer's guide, but I need a better one)" . Lawrence noted that "date and times are all country specific" .
13. GNUe Documentation and Developer's Guide
24�Oct�2002�-�27�Oct�2002�Archive Link: "[IRC] 25 Oct 2002"
People: Derek Neighbors,�Jason Cater,�Jan Ischebeck,�Nicholas Lee,�Daniel Baumann,�Lawrence
Jason Cater (jcater) pointed to the PDF (http://www.gnuenterprise.org/~jcater/Developers-Guide.pdf) version of the Developer's Guide - the source was in CVS, but he asked people not to submit any changes yet as it was a binary file (OpenOffice .sxw) and he had uncommitted changes, which CVS would not be able to merge properly. Derek Neighbors (revDeke) asked "did you get the link i sent about filter to always use uncompressed xml" when saving OpenOffice files, so that they could be maintained properly in CVS? Jason said "yeah - I need to look at that" - he had not had an opportunity to yet.
Derek dug up an old document on GNUe release procedures (http://gnue.org/~dneighbo/gnue/release_notes/gnuereleaseprocedures.html) which he would "like to revamp" and "make it so we can hand a release checkoff to non developers" . Jason suggested a revised version involving lots of vodka, setup.py files, rum, the mcmillan packager, headache tablets and tequila. Derek said "rofl - no wonder we never release - its too expensive :)" .
Jan Ischebeck (siesel) noted that "the new openoffice developer version should support docbook import+export" . Jason said "not really - they say that - but that's what I'm using - and it doesn't have it. I think you need Java 1.4 installed - plus a few other things" .
Jan also asked "what about the next release? will it be 0.4.1 (i.e. bugfixing) or a 0.5.0 ?" Jason suggested "if I get curses working well in short time 0.5.0 - if not, 0.4.1" . Also, "I don't want to go 0.5.0 without a good developer's guide finished - well, finished is a bad word - usable :)" Jan said "btw. the developers guide is great. finally I know how to add timestamps, etc. :)"
Derek "STRONGLY warns all those silly people thinking this developers guide is a good thing - that jamest and jcater plot to torture users - you see if you make a really great user guide and release it - and people find it useful - then when you release 0.5.0 and half the implementation is different - it makes for great luser torture :)" . Jan agreed - "that's the problem with documentation: you have to update it. :)"
Derek reported some problems with spacing when printing out the Developer's Guide from a non-free PDF viewer - all he had available at his current location. Jason said there appeared to be no problem with "acroreader" and Lawrence confirmed it was fine with ghostview. However, Jason warned "I'm having to use ps2pdf (after printing to PS within openoffice) - and ps2pdf is known for crappy, crappy font handling" . "OpenOffice 643 has an "Export to PDF" menu option - but she segfaults every single time I try it - so as soon as OO gets that working - I'm sure it'll be irrelevant" .
Some days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.28Oct2002) , Jason said he had put an "updated developer's guide at http://www.gnuenterprise.org/~jcater/Developers-Guide.pdf (http://www.gnuenterprise.org/~jcater/Developers-Guide.pdf) (for those reviewing it) - I'm at 598 pages now" before correcting this to "58 - maybe 598 if you try to print the PS to a non-PS printer" . Nicholas Lee (esands) said "damn 598 pages would be a book. So who's going to write the GNue book? I hear if you are an author with Oreilly you get all their other books for free. ;)" . Jason said he "expects the Developer's Intro to be over 200 pages" by the time he had finished - "not that I'm counting pages or anything :)" . He was especially pleased that "the entire Appendix B was created with a python script which reads the stuff in GFParser.py" and automatically documented it. Daniel Baumann (chillywilly) asked "what format does OO use, just it's own thing?" Jason said "yeah, it's own XML format" . Daniel asked if "it lets you save as pdf?" Jason said "no, but ps2pdf does :)" The latest version of Open Office "actually has an Export As PDF" but "it doesn't work - but that means they plan to have it on the next stable release."
Later, Derek Neighbors (derek) said "in all seriousness a gnue book is in the works not from oreilly - oreilly doesnt match our principles - we deal with a real publisher that respects the GFDL :)" However, "until its more official we generally dont discuss - i believe our very own peter sullivan will be the primary author" .
14. Performance and Overhead issues with AppServer
26�Oct�2002 (2 posts) Archive Link: "[Gnue-dev] Appserver Interface Issues"
Topics: Application Server
People: Jan Ischebeck,�Reinhard M�ller
Jan Ischebeck said that "While writing a gnue forms client in javascript (http://www.gnuenterprise.org/~jan/jsclient/gnue-forms.html) " , as discussed in Issue�#52, Section�#22� (20�Oct�2002:�Javascript forms client) , he had "discovered some big drawbacks in the forms <-> appserver communication."
"1. PERFORMANCE: for a simple form with 4 entries, 10-12 rpc calls have to be executed just for initialisation and loading the first record. Because most of the rpc calls have to be called sequentaly, the whole initialisation takes ca. 10 sec on a LOCAL connection. Loading a new record takes 5 rpc calls." Reinhard M�ller was "quite surprised that RPC calls are so expensive (performance wise). Doesn't that mean that a single RPC call takes a whole second on a local connection??" If this was generally true, "I'm not sure whether XMLRPC is usable at all for our case." .
Jan also mentioned "2. OVERHEAD+UNCONVINIENCE: In the actual implementation we use a list of instances on the appserver side. But this list is not really navigatable, we just can fetch one element after the other. Because of this, the whole list has to be stored on client side too. and prev, next functionality has to provided on client side. It would be good to define the role of the geasList class more concrete. So we can decide if we keep it as lean as now, or if we add some navigation functionality. Another option is to add the navigation functionality to the instances. a geasList could also be similar to java iterator. ..." Reinhard was not familiar with Java, "but I would strongly recommend to put the needed navigation functionality into the server. The server is the part that should handle caching and all that stuff, so navigation _has_ to be there IMHO." .
To resolve the performace problem, Jan suggested combining some calls, such as "reqeuestNewSession + login(user,pass)" or "createList(table) + setPrefetch() + setConditions + setSort + populate" . Reinhard said he had already considered something like this, and "I think it will be good to rediscuss the API at our meeting" , as discussed in Issue�#52, Section�#24� (21�Oct�2002:�GNUe Developers meeting in Germany) . He stated "My goals for the API are that it's usable, complete, systematic, easy understandable and not bloated." Jan also suggested providing "functions to load more then one record, and to load all fields of a record" . Reinhard said "I'm not sure about loading more than one Instance at a time. If calls to the server are _really_ that expensive, I believe we have to make them cheaper anyway. Definitively, I'm against getAllFields. This will _kill_ performance because we most probably don't need all fields of an object, and the unneeded fields might be long texts or even blobs with pictures and the likes." . But all of this could be discussed further in Frankfurt.
15. Designer usability and architecture
26�Oct�2002�-�27�Oct�2002 (4 posts) Archive Link: "[Gnue-dev] Issues trying to learn how to use GNUe tools"
Topics: Designer
People: Stan Klein,�Robert Jenkins,�Jason Cater
Stan Klein reported several issues trying to learn how to use the GNUe Tools, including "The only way I have been able to get Designer to work is using the wizard." Trying to place widgets directly or via the menus did not seem to work. He had read both the documentation and parts of the source code to resolve some of his problems, but was not much further forward. "If someone can help me with understanding these issues, I will try to find the time to help improve the relevant documentation." .
Robert Jenkins said that Designer "seems to have a lot of capabilities that are not obvious, but also a few missing functions (or at least I can't find ways of making some things work)." He said "If you click a button on the toolbar, you then need to click & drag on the form to set the location and size of the widget. It seems that you can directly drag widgets about after placement, either individually or in groups, but not re-size by dragging so far. Some things such as blocks don't seem to be visible, but exist and have properties. Others such as dropdowns don't seem to work yet. The only way I've found of editing some items is to select 'object navigator' from the Tools menu, then double-click an item in that to get it's properties. The object navigator gives a tree view of the complete form structure with all items on it, & is an excellent feature!"
Stan tried dragging in both 0.4.0 and CVS versions of GNUe, and reported that this "worked most of the time in both versions." He had been able to get a dialog box to set up a static datasource, "but it didn't create it properly in the gfd file." "I also tried dragging widgets to move them. I found that the moving was very inexact" - it was easier to use "the controls at the bottom of the Designer window." He had also tried the object navigator and properties window, but these seemed to be read-only at the moment. He noted "One other thing I noticed is that the Designer dialogues all look similar to the wizard dialogues based on the templates. If this is true, it might provide a means for spreading some of the gruntwork in Designer development."
Jason Cater advised that Static datasources were "very, very new. They work in CVS, but did not work in 0.4.0. The Combo Box wizard that was in 0.4.0 incorrectly had that as an option... that was an oversight on my part." "Now that static datasources are a reality, I need to go back and enable this in the wizard, and properly code it to create them." He agreed that the "dragging" of widgets was not ideal - "This is one of those "I wish wx would do ...." situations. This was the best we could get wx to do up until now and we thought it better than nothing." However, "the WX team has frozen the 2.3 development branch in anticipation of a 2.4 stable release within the next few weeks." GNUe was currently using the 2.2, the last stable release. "we will give the major distros a little time to upgrade their packages, then we will switch to their 2.4 API. There are quite a few additions/improvements that will help us tremendously."
Jason also explained that the similarity between the Designer dialogs and the wizards was "not by chance. The major change in Designer that bumped it up to 0.4.0 was the move to a plug-in architecture." Even the basic dialogs were now implemented as plug-ins. "The beauty of this system is very little about the internals of Designer is needed to write plugins, so practically anyone who understands forms and basic Python programming can do one." They could even replace existing dialogs/wizards with their own, without "modifying any of Designer's source code." However, "Since this was the first release under the new plugin system, we still need quite a few important plugins. Notably missing are AddButton, AddCheckbox, etc." Plug-ins were not very well documented yet, but "There are some technotes in designer/doc/technotes/" He concluded "All in all, the 0.4.0 release of Designer was a stepping stone from being a "hack" to being an all-purpose design tool for GNUe. Consider 0.4.0 to be "puberty" for Designer. Obviously it has a lot of polishing ahead of it, but I think we have the "core" exactly where we need it to move forward."
16. SAP-DB as a free (GPL) database back-end to GNUe
25�Oct�2002�Archive Link: "[IRC] 26 Oct 2002"
Topics: Why GNUe?
People: Lawrence,�Derek Neighbors,�Charles Rouzer,�Peter Sullivan,�Jason Cater
Lawrence (lawrence) felt that, in Jason Cater's goat (http://www.gnuenterprise.org/~jcater/goats/anothergoat.jpg) cartoon, the "goat needs a hat, holster and gun - his name would be Wyatt Erp, background will need $AP, People $oft tombstones" . Derek Neighbors (derek) said "i would agree but i sadly have respect for sap now :( - such a shame - but doing a SAP-DB as GPL and the amount they are putting into the community in that area just blows me away" . Charles Rouzer (Mr_You) suggested "I'm sure they would love the market share like Apache has." Derek said "right now i would rather see navision, peachtree, quickbooks, greatplains - in the background as we will replace them first" . Lawrence felt that SAP's decision to place SAP-DB under the GNU General Public License (GPL) had "ulterior motives -- consulting" . Derek said "no doubt they have motives and im not saying i stand behind r/3 or mysap - just saying i have some respect for them, which i didnt before getting involved with sap-db" . Peter Sullivan (psu) said "in a way, people like SAP grokking free s/w is more important than volunteer projects like this - as in "more joy in heaven over one sinner repenteth" etc - of course, w/o the volunteer projects, free s/w would never have even got started" . Derek agreed - "that is my 'amazement' - that they chose the GPL instead of making the SAP Open License - some lawyer somewhere at SAP gets it. I am also amazed that they have about 10 people full time just answering community questions and interfacing with community for SAP-DB and about 100 person team total for SAP-DB" .
17. Sales Tax in the USA
27�Oct�2002�-�29�Oct�2002 (6 posts) Archive Link: "Re: Can someone explain how sales tax works in the US."
Topics: Sales, Financials (Accounting)
People: Lawrence,�Stan Klein,�Todd Boyle
Lawrence said "A general sales tax solution in US is a very complicated thing. Best to have it configurable." Some states allowed taxes to be set "by what every geographic area suits their ends -- usually by county." He pointed to a link "to read more (http://www.vertexinc.com/) about US taxes" from a company that provided a software solution, using "geocodes" based on US Postal Zip Codes. "Even though geocodes and zipcodes are not an exact mapping of tax rates to a tribes defined taxing area, tribe elders are happy when you send them tax money. For they themselves cannot figure out how to verify that taxes are being collected accurately." Inevitably, some people tried tax avoidance by moving "their postoffice boxes to a lesser taxed area." He warned "This information may be quite dated as I have not looked into this for a while." He also pointed to a very good detailed discussion on USENET (in the comp.databases.ms-access newsgroup) which had been archived on Google Groups.
Stan Klein said "There is also the issue of what is taxable" , giving several examples from states local to him, including "A few years ago, Maryland had a one-time "sales tax holiday" on clothing and school supplies the month before school started. What is taxable can change any time the Legislature meets." However, "an enterprise does not need to collect sales taxes unless there is a legal "nexus" between the enterprise and the state. I think that means the enterprise has to have a bricks-and-mortar presence in the state." There were also excise taxes on goods like liquor, and other local taxes on services like entertainment and hotels. Local taxes would often not co-incide with Zip code boundaries, especially since some municipalities had specific purpose taxes (such as business improvement) that only applied to part of their area. Enterprises with employees might also have to deal with local income tax "piggybacked" onto the state or federal income tax.
Todd Boyle felt that the "automation of sales taxes are now held up by at least two fundamental problems." The first was the requirement for "very big geographical/mapping technology, i.e. expensvie and commercial" to map physical location to tax regime. Also, there was not even "any standardization of the semantics for naming or describing a tax or its tax rate, the bases of what is taxable and the bases and units of measure used for taxation (by the gallon, by value, etc.)" He felt this complex situation suited only "commercial vendors who sell solutions in this market" . Any attempts by politicians to simplify, automate or standardise this was opposed by those who imagined "that by blocking funding for the IRS or anything associated with automation of transaction systems, they might someday reduce the power of the state. sigh. Instead all they achieve is concentration of power in financial services corporations and software companies, and wasting the time of millions of workers in the economy..." .
Lawrence said that "Some enterprises use a zipcode for geo classification. Though not accurate -- maybe not even legal. But so far it has been considered good enough. As long as the taxing jurisdictions get some tax money. In some cases questioning where or not taxes are collected accurately is not technically feasible. Consider, a telco like WCom. One of their billing systems runs at 4,000,000+ sales transactions per second. A phone call may have a "sales tax". Who has the resources to determine where or not the billing system is calculating taxes to the letter?" .
Stan felt that Todd was overstating his case somewhat, but said this was "an issue that has been around for many years but only became important because of the Internet." Previously, only "small, mail-order companies that sold niche products by catalog or magazine advertisement" had avoided sales tax - larger catalogue firms "also had brick-and-mortar presence, so they collected the sales tax on each order" . But "The Internet and e-commerce have created what is essentially an enormous number of catalog merchants" , and this was beginning to have an impact on sales tax revenues for states and localities, who had been lobbying Congress about this. "A few years ago, we discussed this issue in the IEEE-USA Committee on Communications and Information Policy. At that time I suggested that it should be possible to create a simplified collection of data that could be used by catalog merchants for collecting state sales/use taxes." "This could reduce the number of variations to perhaps a few hundred, which is manageable for computing purposes. If a state wanted to participate and have its use taxes collected, it would have to modify its sales/use tax structure to comply with the standard." He had been "told that there was a proposal similar to what I had described either being discussed" , but had not "heard anything about it since" . Although "Congress would need to be involved because of the interstate commerce aspects" , the onus was ultimately on the states "to compromise out their individual ideosyncrasies to get a usable standard."
18. Forms Guide for Develeopers
27�Oct�2002 (1 post) Archive Link: "Comment on Forms Developers Guide"
People: Stan Klein
Further to Issue�#52, Section�#23� (20�Oct�2002:�New Developer's Guide to Forms) , Stan Klein offered "some comments on the draft Developers Guide" . He felt that the form tags list needed to be up-to-date - it "looks to me like they are intended to go into Appendix B." He also felt some general SQL linking discussion would help people's understanding of master-detail relationships - "You did some of this in the discussion of conditions, and it helped" . He would also like some information early in the guide on "how to configure a button to bring up another form" as an alternative to the drop-down. "Of course, someone could also get there using Navigator. :-)" .
19. Security framework for GNUe
27�Oct�2002 (1 post) Archive Link: "Planning a GNUe Security Framework proposal update"
People: Stan Klein
Stan Klein had looked "at the GNUe Security Framework proposal that I last updated in May 2001. I think most of the ideas there are still valid, mainly (1) that GNUe can't provide any security by itself but needs to use the security capabilities of the operating system and database on which it runs, and (2) that different enterprises will have different security requirements that GNUe will need to be able to accommodate." He was now going to re-work the document and re-release it for further comments.
20. GNUe Developers meeting in Germany
26�Oct�2002�Archive Link: "[IRC] 27 Oct 2002"
Topics: Application Server
People: Christian Selig
Further to Issue�#52, Section�#24� (21�Oct�2002:�GNUe Developers meeting in Germany) , Christian Selig (lupo) confirmed that the "european GNUe devel meeting is organized - there will be a separate room, sponsored by reinhard - jan and I will bring laptops with us and keep everyone on track via IRC" , unless "we have no net access, which I doubt). I will try to make some dia drawings when it makes sense. Main topic will be appserver design. I think that's the piece of GNUe which currently needs the most work. When I understand appserver, I can try to write a crash-course for developers who want to use it, because it was an opaque technology to me" .
21. n-tier vs. 2-tier - when not to use Application Server?
26�Oct�2002�Archive Link: "[IRC] 27 Oct 2002"
Topics: Application Server, Why GNUe?
People: Charles Rouzer,�Andrew Mitchell,�Jan Ischebeck,�Jason Cater,�Daniel Baumann
Thinking further about some of the issues from Issue�#52, Section�#22� (20�Oct�2002:�Javascript forms client) , Charles Rouzer (Mr_You) asked "does trigger code get executed on clients and appserver or both?" . Andrew Mitchell (ajmitch) said "mostly clients, depends on the form - not sure how well appserver does triggers at the moment, would have to ask siesel or reinhard" . Charles noted that Jan Ischebeck (siesel) had said that "the important trigger/business logic/consitence checking etc. should be done in appserver, in case of javascript AND python clients." Andrew agreed - "so that'd be remote triggers in appserver - there might be some stub code for calling that within the forms - i am not certain - just requires a simple xml-rpc call, i guess." However, "it depends on the forms - most forms now have inline python triggers - but hopefully they'll move away from that" . Charles said that "the best route for the javascript client will be XMLRPC only" .
Jason Cater (jcater) said "that's going to be sooo damn much overhead - having remote-only triggers - can you imagine the communications between client and server for every single keystroke, etc" . Charles said "X has high overhead but people still use it ;-)" . Jason said that, where an X server was shared among many clients, it was normal to have multiple/back-up servers to avoid "a SINGLE point of failure" . Charles asked "so how do you propose high avaibility appserver? ;-)" Jason said "why do you assume I'd ever use appserver? I think it's great people are wanting it and writing it - but that doesn't translate into me ever using it" . Using an application server simply added another point of failure to the system. The same argument could in extremis be applied to the database server as well, but database servers added more value and there were hardware and software solutions for high-availablity of database servers already. He "went the route of having middleware and my" High Availability (HA) "went <poop> :)" . He was "not thrilled about GNUe requiring an appserver for all the accounting stuff - not thrilled at all but that's not my call" . Charles said "ahh well you just use triggers for business logic?" Jason felt this was part of what triggers were for in the first place. Charles said that "appserver could provide HA capabilities at some point" , but that was not the point - high-availability "is expensive and not normally available to small biz" , so avoiding Application Server to maintain HA was unnecessary. Jason said "for a small business, you are adding yet one more layer that can, and at some point will, go wrong - /me works for a small biz - it's a statistics game - the more layers you add that can go down, the more often you go down. In my mind - until you have multiple database servers - something like an application server isn't gaining you anyhing but rather costing you" . He felt that the "current trend toward middleware" was not "looking at historical evidence and case studies. For some, middleware makes sense. But I see more times than not these days, "if it's not N-tier, how on earth am I supposed to use it" - Enterprises were successfully doing 2-tier (and even 1-tier) long before n-tier" .
Daniel Baumann (chillywilly) said "appserver should only have object/business rules in it not "triggers" as in ui events imho" - "triggers for forms is not necessarily methods/triggers for the appserver - as I agree with jcater you don't want frelling rpc calls for every little ui interaction" . "then again how often does a form hit the db? not on every single trigger does it? I don't think it works that way..." .
22. Mailing list for DCL tickets
28�Oct�2002 (3 posts) Archive Link: "[Gnue-dev] Mailing list for DCL tickets"
Topics: DCL
People: Jan Ischebeck,�Derek Neighbors,�Jason Cater
Jan Ischebeck proposed "a seperate mailing list for new tickets" in DCL, GNUe's bug and task tracking system. This would provide "Notification about new tickets for NON dcl users" , allow tickets to be "browsable/searchable by google ( through mailing list archives)" and provide an "easy way to link to tickets from the web" . Derek was not convinced - "Anyone that wants to watch tickets can email me and I will set them up a DCL account and they can put watches on what they like. SOON, DCL will notify the 'contact' of a ticket on all work done for their item. Regardless if they are a DCL user." Also, "The tickets should be readily searchable already with the guest account." The ability to click on a DCL URL in a notification e-mail to log in and see the ticket had already been suggested in Issue�#47, Section�#18� (15�Sep�2002:�Opening referred tickets in DCL) - it had been added to the CVS version of DCL and would be in the next official release. Overall, "If others think this is a good idea we can certainly do it. I am not sold 100%, but if others find it useful, I cant say I think its damaging." Jason Cater was also not sure, although he had assumed people were receiving notifications from DCL when tickets were closed anyway. He wondered "if DCL should support native "guest" accounts." "Anyone clicking on a direct link into a DCL ticket, if they weren't already logged in, would be taken to the ticket, but as the "guest" user." Installtions that did not want a guest user could just not set one up.
23. Parsing error on Forms without any fields
27�Oct�2002�Archive Link: "[IRC] 28 Oct 2002"
Topics: Forms
People: Bajusz Tam�s
A problem was reporting when trying to build a very simple GNUe Form Definition with just two labels - this generated an error message 'There are no navigable widgets in this form. Unable to display.' Bajusz Tam�s (btami) said "it's a known error IIRC" - "try to add some field" . The way that the Forms parser currently worked, a form required at least one field (as opposed to labels) in order to parse. "it was ok before, but things changed :) i don't know the real reason" .
24. GNUe on hand-held PDAs and python versions required
27�Oct�2002�Archive Link: "[IRC] 28 Oct 2002"
People: Jan Ischebeck,�Jason Cater,�James Thompson,�Derek Neighbors
Jan Ischebeck (siesel) said "does anybody knows if gnue-forms runs on python1.5 too? Or why it doesn't work anymore." Jason Cater (jcater) said "it will not - it requires 2.0 - we use 2.0 features such as +=" . Jan said he needed to get "gnue running on a WIN CE 2.11 device - and python2.2 for CE just runs on CE 3.0 :( The third choice would be using WABA (java for palmtops) but the jclient is very very preliminary" . Jason said he was not keen on reviving the Java Forms client. James Thompson (jamest) said "i don't think going back to 1.5.2 is very feasible - we had quite a few 1.5.2 hacks that eventually drove us to 2.0 - putting those hacks back in would be ugly" . Jan emphasised that he was not proposing "changing cvs, just thought of a private patch :)" . James still thought "it would be rough - very rough - we had to do wierd class hacks to get cetain things to function" . Jan asked which earlier versions of GNUe Forms had supported python 1.5 - "so the switch to 2.0 was short before gnue forms 0.3.0, I think" .
Derek Neighbors (revDeke) asked Jan why he was restricted to Windows CE 2.2. Jan said "I have no money for a new PDA :(" . Derek "thinks perhaps you need to flash it and install something new on it :)" He explained "gnue should be focused on architectures to where its going more than where it is - that was a large part of the python 2.x decision. We figured by the time GNUe 1.0 ships - 2.x will be OLD - and 1.5.2 probably about unsupported" . Jan said "if I get forms 0.1.0 working, its ok. The main aim is to present a working gnue on a PDA to a customer. If I get the project, they can use a Linux PDA. i.e. no porting needed." Derek said "python should work no problem - but we will likely need another UI client - qtopia more than likely - or something similar" . He would recommend "going to #handhelds.org and seeing if anyone there has any tips/hints suggestions to getting linux in some form working" on Jan's existing PDA - this might be easier than re-writing GNUe. Also, "what about curses version of gnue? give another 3 months and we can probably proof it on an iPAQ for you" .
Derek discovered that "familiarOS for iPAQ has Integrated Python v2.0 w/ PyGtk and PyGDKImlib. - time to test out that gnue pygtk driver :)" Jason pointed out that the GNUe UI driver was for GTK2, not GTK, but Jan said "it can be easily modified to work with GTK. - but Its possibly better to just install debian unstable on the iPAQ" . Jason agreed - "otherwise, how else do you get frozenbubbles easily?" Derek noted that "familiar provides 100% binary compatiablity to debian dpkg w/ ipkg - er the arm distribution - so in theroy if arm supports pygtk2" .
He was very enthusiastic about the prospect of getting "designer to run on an ipaq - /me can see techie meetings.... no more need to fall alseep one can design forms while in meetings, literally can have things DONE before you leave a meeting :) - instead of saying... "that would only take five minutes" and end of meeting give it to them and say "that only took 5 minutes" ;)" . Jason asked about the screen resolution, and "cringes at the thought of designing w/in 320x200 :)" . Derek said "if you used 'virtual desktops' and a tabbed windows manager (like fluxbox) - it might be doable if you were designing forms for the iPAQ" .
He had "tried VERY hard to get some zaurus' on loan for development but sharp was really bad about it - stating they are only 200 thats cheap enough to 'buy'. i kept stating this is for FREE not PROPRIETARY software - and that likely more FREE software would only HELP Zaurus sales not HURT them - but the marketing drones i had to deal with basically were clueless :( HP/COMPAQ on the other hand is much more savvy - and understands the concept of symbiotic relationships - which is most odd as it should really be reversed seeing how zaurus is the linux pda and ipaq the pocketpc one.... go figure :)"
25. Free/Open-Source ERP portal sites
27�Oct�2002�Archive Link: "[IRC] 28 Oct 2002"
Topics: Why GNUe?
People: Keith Jagrs,�Derek Neighbors,�Keith
Keith Jagrs (KeithJagrs) asked "have you checked out erp5.org (http://www.erp5.org) ?" Derek Neighbors (derek) said it "is what i had hoped the gnue.org community site could be like" . He explained "at one time i wanted to do somethign similar to erp5 - as it was needed - and we caught major crap for posting non gnue stuff - and it was confusing - it really needed to be a separate site" . This was "our old phpnuke site for gnue.org - i had hoped to do more an ERP community tyep of thing - but we needed to make it a second site - as to not confuse people - i.e. i agree with decisions we have made - and honestly we dont have resources to do a community site like that plus our project" .
Keith asked "taking its site aside, or its community, what do you think of their project, or their technical approach?" Derek said "i only saw it as an erp portal - and them trying to collect best of breed together. btw: this is one of the things gnue was supposed to do - compile stuff together in a sense - but i digress" . He "didnt see that they were actively 'coding' anything - but rather consolidating/packaging existing things - i could be wrong i only looked for 10 seconds" . Keith said "it seems to me a peer project - they are based on Zope as the core for the app server" .
26. GNUe small business packages
27�Oct�2002�-�28�Oct�2002�Archive Link: "[IRC] 28 Oct 2002"
Topics: Financials (Accounting), Customer Relations, Manufacturing, Sales, Supply Chain, Document Store
People: Derek Neighbors,�Jason Cater,�Peter Sullivan,�Nicholas Lee,�Andrew Mitchell
Derek Neighbors (derek) asked "where to put packages :) for gnue" . He had to get some work for a client started "asap - i have the stuff in private cvs - i would like to address what to do with that stuff - as well as get mroe information on gear stuff" . He wanted "to spend minimal time talking (yeah amazing eh) and roll up sleeves and get a hacking - but im to point i need good place to put this stuff as im going to encourage others to contribute ASAP. acclite fits in that mix as well - a little clean up and really i think the base of 'stuff' is done - i.e. product categories and product maintenance - which is the start of ordering/inventory/invoicing etc - i.e. partnumbers woot!" . He "was going to consider revamping the module writers guide or such - and start creating two tier module docs - and merging in louis accounting stuff and comparing to nola and such - /me is gung ho - ready for some public announcements and website updates and code committing" . His main question was "mainly where to put the work" . Jason said "I'd like to see this be the start of using separate modules - i.e., I'd like to see gnue-accounting - (or whatever you want to call it)"
Derek said "long term i want to know where we put on website (as discussed) - and package management. i would like to see this be the start of packages on the website i.e. official packages so to speak. I am down with modules but it gets shaky on data dictionaries - as if i make say CRM and Inovicing - both use the contact database - so where do i put contact data dictionary stuff? If i put in crm then invoicing now depends on crm - /me doesnt expect easy answer here or a instant solution as i think its MUCH more complicated than that" . Jason said "um - just put it somewhere - we'll move it around later :)" Derek agreed - "i wouldnt be opposed to saying X depends on Y for NOW - as having usable stuff will make folks so happy it shouldnt matter much - as long as we have an eye for a better solution long term. i also considered putting data dictionary all in one place and loading all tables regardless of what you install - but then you only have to get the forms, reports etc of what you want" . Jason agreed.
Jason asked "what is the official stance on what you're doing? these are stepping stone packages? thesse are unofficial, but working solutions?" . Derek said he did not mind - "you know what i have to make - about 3 or 4 folks are wanting identical thing - whether we make this an unofficial soslution using the framework and put in speical cvs and special website - or whether we put on gnue as official 'starter' packages to be replaced later - or whether we say these are official packages that will grow over time - makes no difference to me. I just want to be able to advertise to solict 'help'" . He thought "making them officially gnue likely will help gnue's image - look at our website, its apparent we want to offer working systems :)" . Jason had no strong views either, "but for political reasons, it may be best to have this as an unofficial solution using the framework - it's a tricky situation :) I can tell you now, it'll be "damned if you do, damned if you don't" - so choose your poison now and move on."
Jason "would probably lean towards "usable, but not quite official" solution in a separate cvs (but on savannah (http://savannah.gnu.org) )" (the FSF's free software alternative to sourceforge) "but still advertise on our website" . Derek wondered "why not just use existing" CVS on GNUe's own server "where we have acclite and/or UE stuff" as of time of writing. Jason said he "would like to see anon cvs" but "neither jamest or I are ready to open up" GNUe's own server "like that :)" . Derek remembered "thats right no pserver there - ok i will start a project then on savannah - make a little web page and such" . Jason said "I would name it gnue-????? and put on gnue website - just be careful with wording :) - so it's "associated with" GNue even if not in our main cvs repos" .
After midnight (http://www.gnuenterprise.org/irc-logs/gnue-public.log.29Oct2002) , Peter Sullivan (psu) said "I would still rather we did the back-office stuff packages + acclite rewrite as "real" GNUe packages if at all possible - even if 0.0.1 or 0.1.0 versions" from "a webmaster/PR point of view" . He felt "people would rather have packages that meet someone's need - and can be used as a basis for something more general - than get stuck in Proposal Hell for another 12-18 months" - if the problem was "just the need to be seen to follow the proposal process - I would say change the process." Nicholas Lee (esands) asked "WHat is the actually plan for acclite at the moment? Current having hacked around the code somewhat it just seems to be a modified version of nola to work _somewhat_ with postgresql. Is someone working on forms for it?" Peter said "I believe that a couple of screens have been converted to GNUe Forms - but that it has been a bit on back burner" until now.
Later on, Derek Neighbors (dneighbo_) asked whether the new CVS for GNUe Applications had been set up on savannah yet. He explained "we have some 'small biz' stuff getting ramped up and ready - there was concern about putting in place of 'official' gnue packaging - and we definitely didnt want in the gnue tools cvs" . He said "i have product management basics ready - and contact management stuff is nearly ready - those two things are the BASE for ordering/invoicing/inventory and the likes - as you have to have product and customers :)" Nicholas Lee (esands) said "Hopefully when I get this" Foreign Exchange (FX) "management stuff sort, its something you might put in there." Derek said "i also plan serious ramp up and extension of pieces of acclite" . Nicholas asked "What is the roadmap for acclite? You going to work with the php base at all? Or develop the form/gnue side of things?" Derek said "originally i had planned on keeping the php side up - but more i look at it its likely i will not - but others could - /me suspects it will depend on how much we start modifying the underlying structures - if its minor tweaks there are reasons to keep the php up if its more major or new stuff its probably not worth the core's time" . Nicholas asked "if the schema is worth using" . Derek said "if the schema isnt worth using we will modify as necessary" . Nicholas said "Main thing I'd need fix in order to use it, is sales tax on the vendors side" . Derek said "i already have own product structure and own contact structure - /me has not compared how that looks to acclite - at first gander i suspect (hope) to rip" Accounts Receivable, Accounts Payable and General Ledger "from acclite - likely might not use anything else" . Nicholas "wonders if its worth using the nola base at all, or starting for scratch" . Derek said "looks like it has some docmanagement stuff - but the ui of forms might not be good for that" . Nicholas said this was "nothing to do with core accounting stuff though. Doesn't dcl handle some doc management stuff? I think though that doc management is an important part of contract/CR management" . Derek said "doc managment has about 10 different sdies - i think ulitmately we will need another 'tool' for this - one that forms can interact with" .
Derek emphasised that "one of the pluses i saw in not doing official gnue is no hold ups - i.e. this isnt design by committee :) muhahaha - input definitely wanted - but things will be moving faster than making the right decision the first time :) When you get 300 people on a mailing list and design software - you get no where quick :) - as there are a million well if you do this you get X - and basically every sanctuary has its price - there is rarely a definitively right answer" . Nicholas agreed - "develop something that works for people. Either i) people will come and improve your work, ii) people will come see your work and have that flash and see how they can complete rewrite to to work for them and you - Either way its good for you and your got something working" . Derek agreed - "just stating that the small biz stuff (because of outside requirements) will likely be very much do, refactor, do, refactor, do - we have 2 or 3 'customers' that are really driving the development - originally we were going to do in a more private cvs and wait until release to open cvs - but we feel that is not the free software way - we would rather let people play interactively - even if sometimes we have to say too bad" as the needs of the paying customer had to take first priority. It was about "baby steps - being able to catalog your products on a computer is better than not being able to even if you dont have inventory of them" . "so one might say products w/o inventory module is dumb - i say products is base for inventory so its progress :)" He felt "the two things i have seen 'drag down' projects - a. bickering to decide on decisions for so long that you could have implemented in all proposed ways in the amount of time you spent arguing about it. - b. not starting anything until you have decided on EVERY possible feature and/or scenario" . Nicholas said "b) never works though. Prefection is impossible to get. Afterall who's prefection is it? Obviously its the person actually writing the code, thus you are back at some form of a) again. Anyway it'll be good to see what you guys have got so far and what your roadmap is. I can provide input on how that would work for the business enviroment here" in New Zealand.
Responding to Derek's original question, Jason Cater (jcater) said he had applied to the savannah-hackers mailing list to set up a new CVS, but had not heard back from them yet. He asked "what's the typical turn-around to requests like mine?" Andrew Mitchell (ajmitch) said "it's a project submission, so could be 2 days if you're _really_ unlucky - usually a few hours, depends who's evaluating" . Derek let off some steam that GNUe had had some problems with support requests from savannah and the FSF generally recently but added "its really not fair to complain after all its volunteers - sometimes i think they could structure better though" .
27. Still some focus problems with Forms
29�Oct�2002 (1 post) Archive Link: "Re: [Gnue-dev] Re: focus nightmare"
Topics: Forms
People: John Lenton
Further to Issue�#52, Section�#15� (20�Oct�2002:�Focus problem with multi-line Forms) , John Lenton reported that "the click-on-the-border-and-watch-the-focus-go-anywhere nightmare is still there -- but that's probably wx." .
28. GNUe Web Site organisation
28�Oct�2002�Archive Link: "[IRC] 29 Oct 2002"
People: Nicholas Lee,�Peter Sullivan
Neatly linking together Issue�#53, Section�#6� (23�Oct�2002:�GNUe Roadmaps and general project organisation) and Issue�#53, Section�#25� (27�Oct�2002:�Free/Open-Source ERP portal sites) , Nicholas Lee (esands) said "I don't really have the server spacem, but I can set a (t)wiki up easily enough. The question might be though, if erp5.org are doing it. Is it worth doing repeating?" . Peter Sullivan (psu) said "if people really want a wiki, I would rather host it as wiki.gnuenterprise.org" but "At the moment, I would say anything that people might want on a wiki e-mail it to" him "as I'm sure I can find a sensible place to slot it" on the main web site.
Peter said "The front page still concerns me - The "What is GNUe?" needs to be up front for newcomers - but for old hands, we really want the news to be first up" . He wondered about using cookies to manipulate this, but Nicholas suggested "Just have two different entry points - People bookmark the page they want. Otherwise the server is making decisions for the client" .
Peter said "The big issues on the website at the moment are - a) Docs need updating - I just need to get around to recompiling them from CVS - b) FAQ needs updating really badly - A lot of the material is already on the website - just needs folding back into FAQ. But then, as always, we accept patches ;-)" . Nicholas said "wikis tend to make that process a little easier. (provision of patches)" . Peter said "Actually, #1 priority is to get website into CVS so other people can touch site w/o" having to let him know what they had done - he would pursue this further.
29. Writing GNUe applications in other languagues
28�Oct�2002�Archive Link: "[IRC] 29 Oct 2002"
People: Bill Gribble,�James Thompson,�Derek Neighbors,�Nick Rusnov,�Jason Cater
Bill Gribble (grib) asked "can anybody here comment on the use of the designer and other tools with Guile? are they totally tied to python or somewhat language neutral?" James Thompson (jamest) said "right now they are tied to python - we have plans to support triggers in other languages - but no code exists yet to take advantage of it" . Derek agreed that GNUe was "tied to python - designer will likely always be unless there is a port (me sees little value in a port) - as to the forms client (forms is a spec, there already are php and javascript and java implemenations in cvs) ableit far from complete. i would say there is LITTLE need to make an implementation in a separate langaguage - as they are TOOLS not applications i.e. we plan to support business logic in ANY language so people writing gnue applications should be able to write applications only knowing their favorite language" . However, "creating an 'alternate' implementation of the tool is a lot of maintainance and duplication (but as stated we have samples in cvs so its doable). Python is about as portable as it gets - i.e. from handhelds to mainframes" . He also noted "there is a native gtk2 version" of Forms "in cvs in case wxpython is one of your concerns" .
Bill said "I am using a similar designer via glade that's tied to scheme in the same way; I just happen to like scheme more :) The nice thing about the gnue designer is the twist you guys put on it to support character cell layout. I am pretty likely to be doing a significant chunk of ERP type stuff for some car dealerships and I was hoping to give/take some with gnue, but I don't think I'm ready to switch the tools over to python." He did not "have anything against python, I just like lispy languages better. guile has its limitations but at least I know what they are." Derek said "guile/scheme looks very cool - problem is it has a big learning curve compared to python for business programmers. I dont think that makes it bad, just we tried to pick something closer to our target market plus larger programmer base" . He suggested "the 'right' thing to do if you wanted to do this and not loose scheme is adopt the formats i.e. keep own designer - just make it write out gfd's and grd's - then make a forms client in scheme. This shouldnt be overly difficult - and it woudl actually give you compatiability to GNUE - i.e. a form you write should be usable by python clients as long as you are in spec - especially once we have cross language trigger support" . He noted "we are getting ready to rollout quite a bit of application base very soon - geared towards manufacturing - but most of it would be relevant to car dealerships too - as keeping books and paying bills is pretty universal :)" .
Nick Rusnov (nickr) said "Seems like the gfds should allow you to say 'this trigger uses this interpreter'" . Jason Cater (jcater) said "we have that functionality, just no sublanguages - all triggers support a language="" attribute - just right now, "python" is the only valid answer :)" . Nick asked "does the forms client really have to be written in the same lanugage as the triggers?" Derek said not - "the idea is that python version will support triggers in ANY language" that "someone writes a wrapper for" .
Derek asked whether "we could make designer work same way for plugins? i.e. could we make the plugin architecture support other languages" ? Jason said "I suppose so - but that's a slightly different audience - /me isn't sure of the benefits - but it is certainly doable" . Derek said that "if we make a concession for business applications to be outside of python (the biz logic/triggers) - would it make sense to offer this flexibiltiy for designer plugins" as well. As Designer evolved, so that "everything almost is done via plugin - we might get more stuff written if we offer to let poeple write in language of choice - or at least python/scheme/perl" .
30. Debian packages for GNUe into sid (unstable)
28�Oct�2002�Archive Link: "[IRC] 29 Oct 2002"
Topics: DCL
People: Nick Rusnov,�Derek Neighbors,�James Thompson,�Jeff Bailey,�Andrew Mitchell,�Jason Cater,�Nicholas Lee
Nick Rusnov (nickr) noted "hmm, all the official gnue packages in debian" , dating back to Issue�#12, Section�#3� (11�Jan�2002:�Debian packages for GNUe) , "have been orphaned" . Derek Neighbors (revDeke) said "good we need a maintainer to submit what we have" from Issue�#50, Section�#9� (4�Oct�2002:�Debian packages for GNUe) "to official :)" . Nick said "I'd be willing to do that if they don't suck." Derek asked "any chance you could review them for 'suckage' - and if they dont suck take over the packages and get what we have uploaded to debian?" Nick said he would "look em over" .
Later, Nicholas Lee (esands) noted that the old packages for gnue-common, gnue-designer and gnue-forms all had outstanding bugs in the Debian bug tracking system. James said people could "add a line to their sources.list and just apt-get the new stuff - deb http://www.gnuenterprise.org/debian woody main" . These were the Debian packages that Jason Cater (jcater) had prepared, based on those for Zope. Jeff Bailey (jbailey) asked "Are they going into unstable at all?" James said "we don't have a clue how to make it happen" . Andrew Mitchell (ajmitch) said "they probably will if you spruce them up :)" . Jeff said "I can make that happen easily enough. Especially since we might start using gnue here. =)" James said "i'd love to see them in" the Debian testing distribution (as of time of writing, sarge) "as well - but I know they have to play nice in unstable for a while" . Jeff said that, at the moment, Debian "unstable is still suffering with the glibc 2.3 transition and the perl transition" , although this should not affect python-based packages like GNUe.
Jason said "whatever you do - PLEASE keep the changes in CVS - I have all my debian files in our cvs structure" . James agreed - "we tried time and time again to get the files and a simple how-to in cvs - but people treated us like we were derek and just ignored us :)" Jason said "we can maintain them - I don't mind that - we just need a sponsor" . James agreed - "we just need someone wise to tell us what hoops to jump thru - personally I'd like to see the maintainer listed as the project itself - as the orphan problem seems to haunt debian projects" . Nicholas said "No rush though to get the debs back in. Can always do maintain it external and pass it to a debian maintainer to commit to the rep." James agreed - "we have external repository now - i set that up about a month ago using jcater's debs" .
Jeff suggested setting up a small mailing list at lists.debian.org to do the bug tracking for the packages. Nicholas suggested "WHy not just debian@gnue.org ?" Derek suggested "if debian bug tracker has to email something - howabout debian-support@gnuenterprise.org - thsi way it will file bugs automagically into DCL - and we can have one 'official' person in charge - but certainly others can put on watches or the likes" . Jeff said "That's probably suboptimal, since then you'd have two bug tracking systems." Derek said he wanted to let people use the Debian bug tracker system "to log the bug - and we just do something clever to have DCL report back to BTS on things - this way our developers have a single system but people can still file bugs via BTS. i.e. im not against BTS - but we need to have bugs in one place for us - and dcl is that place" . Jeff said it was normally "Debian Developpers job to handle bugs and file them upstream when needed." Derek said "well i would consider this no different - just our debian developer is a 'collective borg' - and so it auto forwards everything upstream :)" since the GNUe project would have committed to also maintain the debian package for itself. Jeff said "Tell you what - I'll upload it at first sending the bugs to me. When you get it wired up, lemme know. =)" .
Jeff reviewed the new packages before uploading them - "it looks like baux orphaned all of the gnue stuff. So it won't be a problem for me to pick it up" . He asked whether "gnue uses zope?" Andrew Mitchell (ajmitch) said "nope - not at all - jcater" just "based the debs off zope's packaging :)" . Jeff pointed out that the README still mentioned Zope. Jason said "the README was one of the things i needed to touch up - as well as our description: fields" . He also wondered "what kind of bribe it'd take to get him (or someone) to get nickr's pysablot debs" , as discussed in Issue�#25, Section�#15� (14�Apr�2002:�Using pysablot as an XML transformation tool with GNUe Reports) , "in unstable? I do know those need to be renamed to python's standard (which I guess would be python2.2-sablotron) ??" Since sablotron/pysablot was a dependancy for GNUe Reports, this "would be ultracool at some point. Or was nickr waiting on something from upstream? I guess we should ask him :)" .
31. Python version for GNUe
28�Oct�2002�Archive Link: "[IRC] 29 Oct 2002"
People: Derek Neighbors,�Jason Cater
Derek Neighbors (derek) reported problems starting Designer - "ImportError: No module named wxPython.wx" . Jason Cater (jcater) asked "did you upgrade to unstable's python?" . The Debian sid (unstable) distribution now used python 2.2 as its standard instead of 2.1 - he suggested that Derek should "apt-get install wxgtk2.3-python" . Derek asked "why didnt the DEPENDENCIES pick up something" - "i.e. why did it switch my python to 2.2 but not fricken grab its dependencies" ? He noted that most python-based packages in Debian used python 2.3 now. Jason said that, according to the python website, 2.3 was not out yet - "and they say debian is behind - bullshit! - unstable is so damn current, it has software that hasn't been written yet!" Derek wondered if someone had packaged python 2.3 from python's CVS rather than an official release.
Jason noted that the plans for python 2.3 included "a lot of changes" - "they have a bool() datatype, generators, enumerators" and other extra functionality that would be useful in GNUe. Derek "wonders what you get if you apt-get install python2.3 :)" Jason said "I would be afraid" . Derek said "if you have to PORT - shouldnt it be 3.0" ? Jason said that the "thing is it's ALWAYS that the new version enforces a language rule that they were lax in before. So the code, once "ported", still works in the older 2.x version :) e.g., 2.2 really enforced namespace scope rules that they spelled out for a long time but didn't enforce - that broke some people's python. But that's because they were writing broken python :) - we don't ever do that <cough>" .
32. Problems with PostgreSQL introspection effect Designer schema wizard
28�Oct�2002�Archive Link: "[IRC] 29 Oct 2002"
People: Derek Neighbors,�Jason Cater
Derek Neighbors (derek) said the "designer schema wizard kicks some arse" . He asked whether the .gsd (GNUe Schema Definition) output file should be "ONE giant schema? or one per table?" Jason said "we've had this debate before" , in Issue�#37, Section�#14� (9�Jul�2002:�Including data in GNUe Application Server schema definitions) . He said "I *think* the outcome is we should support some kidn of <import> mechanism" . Derek agreed - "if we coudl 'include' (or import)" then "single was probably better" .
Derek said he was "SO impressed to see you could pick table listings - only thing missing is it needs to pop a dialog when done - saying 'done' - i just noticed that the term window stopped spitting out debug messages :)" Jason said that this would not be an issue "once we have an editor :)" . Derek noted that the wizard had translated a postgreSQL field of type numeric(12,2) to " <field name="amount" size="786438" type="number"/>" . Jason suggested "somethings screwy with the introspection" . Derek did some poking around inside PostgreSQL and discovered that the data dictionary had a field "atttypmod" which contained the mystery 786438 value - obviously a look-up to another table of valid numeric types - "guess we need to learn 'conversion magic'" . Jason said "ah, crap... we are obviously misinterpreting what atttypmod means" - "postgres catalog table doesn't store like I expected" . Derek was convinced he was "a walking light bulb (fricken bugs are attracted to me)" .
33. Free/Open-Source Financials Packages
29�Oct�2002�Archive Link: "[IRC] 30 Oct 2002"
Topics: Financials (Accounting)
People: Derek Neighbors,�Nicholas Lee
Nicholas Lee (esands) pointed to Gnu Ledger (http://www.mustardandrelish.com/ledger/) . Derek Neighbors (derek) said "you seem to research this stuff like mad - i used to but ran out of time - ANY chance as you find this stuff you could make a database/spreadsheet/textfile that has projectname - category - maincontact - maincontact email" Nicholas said "Sure, I could probably do something like that. However freshmeats (http://freshmeat.net/browse/76/?topic_id=76) pretty good" . Derek explained "one thing i used to do is contact the projects - if it was something that was complimentary i invite them to work with us" - "we at one time consolidated about 10 projects into us" .
Nicholas said "This is an interesting under the radar project I just discovered there: http://www.redrocketconsortium.com/zbc/index.html (http://www.redrocketconsortium.com/zbc/index.html) " . Derek said "i have see the redrocket one before" - he noted it was written in ADA. Ndicholas said it "Has some nice screenshots. Although it looks like it suffers from the same us-centric issue of only dealing with sales tax on sales" . Derek felt that "the problem with ADA is finding otehr developers will be murder - if this guy decides to stop finding someone to maintain it could be impossible or (costly)" . Nicholas suggested "Probably defense industry retirees" .
Nicholas said "Anyway, of all the oss/free-type accounting projects I've looked at so far. None of them are quite there, in terms of being solid mature and robust." "The only three I've keep coming back to are sql-ledger, nola and gnucash (cvs sb version) - Nothing else I've tried is still on my machine. Nola/acclite is still there only because you guys are working on it. Given its broken for the NZ tax enviroment its not very function for me" until this was fixed.
�
�
�
�
�
�
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 kernel.org. All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License version 2.0. |
