GNUe Traffic #24 For 12�Apr�2002

By Peter Sullivan

"this is NZ - in norway they do IP over pigeons, here we have IP over sheep"

Table Of Contents

Introduction

This Cousin covers the three main mailing lists for the GNU Enterprise project, gnue, gnue-dev and gnue-announce. For more information about GNUe, see their home page at http://www.gnuenterprise.org (http://www.gnuenterprise.org) . Details of the mailing lists can be found at http://mail.gnu.org/mailman/listinfo/gnue (http://mail.gnu.org/mailman/listinfo/gnue) , http://mail.gnu.org/mailman/listinfo/gnue-dev (http://mail.gnu.org/mailman/listinfo/gnue-dev) , http://mail.gnu.org/mailman/listinfo/gnue-announce (http://mail.gnu.org/mailman/listinfo/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 irc.openprojects.net:6667, or you can review the logs at http://www.gnuenterprise.org/irc-logs/ (http://www.gnuenterprise.org/irc-logs/) .

1. GNUe at Linux World Expo SF

5�Apr�2002�-�9�Apr�2002 (1 post) Archive Link: "Linux World Expo SF -- interested?"

Topics: Why GNUe?, Bayonne

People: Mark Miller,�Derek Neighbors,�David Sugar

Mark Miller said "I've heard GNUe has been shown at a few user groups with very positive feedback. Would anyone be interested in putting together a demo that could be showcased in the AMD booth at Linux World Expo in San Francisco?"

Some days later on IRC (http://www.gnuenterprise.org/irc-logs/gnue-public.log.08Apr2002) , Derek Neighbors (dneighbo) asked "is there anyone willing to come to Linux World Expo in August" ? He said "it sounds as though AMD will give us facilities to present GNUe for 20 min every hour or so on their floor - which means if we want our own booth we will need more than one person there - i am hoping to be able to make the trek" .

Two days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.10Apr2002) , Derek asked David Sugar (dyfet) about Bayonne's experiences "with being at Intel booth in NY that one year?" . He "was wondering if you had horror stories or not :)" David said "The problem was we were the only free software at the booth...and nobody thought to look for us there" . Derek said "we would maintian our own booth still" . David felt "the other problem with the Intel booth was it was huge and we were so easily lost in it :). AMD might be much better in that respect" .

2. Debian packages for GNUe and DCL

3�Apr�2002�-�4�Apr�2002�Archive Link: "[IRC] 04 Apr 2002"

Topics: DCL

People: Nick Rusnov,�Jeff Bailey,�Derek Neighbors,�Andrew Mitchell

Continuing Issue�#21, Section�#19� (20�Mar�2002:�Debian packages for GNUe) , Nick Rusnov (nickr) said "I'm looking at the current debian packages [...] they are rather wonky" . He noted "the guy put the ./debian directory right into the setup.py - rather than doing the DESTDIR thing you're supposed ot do - I have to reread the quick guide to packaging to figure out the best way to do it. I think it'll involve wrapping setpu.py in a Makefile - althigh since I know the people who wrote setup.py I could just tell them to support debian" .

Later, continuing Issue�#23, Section�#11� (30�Mar�2002:�Debian packages for DCL) , Jeff Bailey (jabiley) said he had "some not-finished debs for DCL." He was "trying to decide if I Should automate database selection and creation, and whether that's an abuse of debconf." . He would be the DCL package maintainer - "It was placed up for adoption and I've announced my intention to adopt it already." He thought "Maybe we should get a debian-gnue@lists.debian.org setup for good group discussion, and do the uploaders thing." Derek Neighbors (derek) suggested "feel free to use the gnue-dev list" . Jeff thought that "The advantage of a Debian list just ofr packaging is visibility." . Also, "while you're different, most upstream folks tend to chew a new asshole in people who report packaging bugs to them." Andrew Mitchell (ajmitch) volunteered to get "a list organised" . Jeff warned it might take some time.

The next day (http://www.gnuenterprise.org/irc-logs/gnue-public.log.05Apr2002) , Jeff noted that "There are apparently only 5 users of dcl" , based on the figures reported by the Debian popularity-contest (http://people.debian.org/~apenwarr/popcon/results.web.html) package. Derek Neighbors (derek) said "there are lots of dcl users (more than 5) BUT i would imagine probably none of them installed from the debian :) - so i dont think 'breaking' the upgrade of the debian is a big deal" .

3. GNUe as an alternative to gnucash

3�Apr�2002�Archive Link: "[IRC] 04 Apr 2002"

Topics: Financials (Accounting)

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

Derek Neighbors (dneighbo) said "i love the gnucash list - rate they are going they will demand smoething like gnue for personal finances :) - i.e. sql database with middle ware via RPC etc etc etc" . He had started using gnucash personally, despite the lack of "budgeting - and the reporting sucks" . Jason Cater (jcater) said "I plan to use GNUe for personal finances? what's the big deal? plus then the wifey can use it on her Win98 machine :)" James Thompson (jamest) had a simpler solution - "i use wife for personal finance - then I don't have to dick with it at all :)" . Jason said "I tried that - but only so many register entries can be written on a woman that small" .

4. Using pysablot as an XML transformation tool with GNUe Reports

3�Apr�2002�-�6�Apr�2002�Archive Link: "[IRC] 04 Apr 2002"

Topics: Reports

People: Derek Neighbors,�Derek Neighors,�Jason Cater,�Nick Rusnov,�Christian Selig,�Daniel Baumann

Derek Neighbors (dneighbo) said he had a real need for a "reporting solution here" at work, and asked how GNUe Reports was progressing after Issue�#22, Section�#1� (19�Mar�2002:�GNUe Reports proposal) - "have you formalized the 'tags' yet for NORMAL reporting? if so im ready to beta test" . Jason Cater (jcater) said "that's what I've been working on last 2 days - I wanna release reports 0.0.1 when we do forms/designer" . He said "fwiw, I'm doing a "generic" tag format that doesn't have to be the "formal" format [...] that will demo what it will do" . Derek said "i will write an xsl for the output to excel - as im hurting here - the report tool sucks so bad here ive dumped the data out to access (from olap) and am using access to write a report and its SHIT. I would rather dump to postgres then use gnue reports" .

Jason "really wants to see an XML to XSLT converter :)" . Derek said "i might try a convertor tonight to html" , using "pysablot" , although he wished there were Debian packages for this. Jason asked "nickr: yes, could you package pysablot.sf.net (http://pysablot.sf.net) ?? - /me would be your best friend - or at least nicer to you than I am to chillywilly ;)" Nick Rusnov (nickr) suggested using "4front instead - which is already packaged" . Jason noted that "sablotron plus the perl and php bindings are already in woody - just no python bindings" . Derek said he had mainly used sablotron, "as its what mdean and i found to have best docs :) - both beign new to xslt this was important :) - we got it working for DCL - and it seems very easy and simple to use - and has a plethora of bindings" . Nick said "I was going to use sablotron at one time but then I found 4front which did everything I needed" . Christian Selig (sledge_) recommended the XSLT FAQ (http://www.dpawson.co.uk) - "it's a faq for those who already use xslt - it gives some good advice on the more tricky issues :-)" .

Two days later (http://www.gnuenterprise.org/irc-logs/gnue-public.log.06Apr2002) , Jason asked "nickr: did we ever volunteer you to package pysablot for us?" Nick replied "you tried to convince me - but I'm whiley" . Jason explained that psyablot was a "a wrapper to sablotron" , which was "an XSL processor fully implemented in C++" . Nick pointed out that "you guys do know that pysablot was last released in nov 2000? and that its still alpha?" He said he would "make a deb of it, but I don't think I should put it in the debian archive if" it wasn't being actively maintained. Derek said that there was a freeze on new packages until the May 1st release of Debian woody (as the new stable distribution) anyway. He didn't think that it was a problem that there had been no more recent releases of psyablot - "i mean all it is a wrapper to sablotron - and from what i have used it works w/o issue - and seemed to have all functionality - so unless sablotron changes i wouldnt expect a lot from pysablot to change - i am willing to contact the maintainer of pysablot to see if they have abandoned it or not" . If so, "i will probably see if he is willing to turn it over - call me sick but i like it :)" . He thought it was important, as "iirc there were only like 2 or 3 xslt packages for python - sablotron and pysablot were the only combo that were BOTH GPL - i also liked that sablotron is maintained by a commercial company that puts money into the development" . Nick pointed out that "python 4suite is free - and maintained" . Derek said "fwiw i suppose in reports and such we could treat it like db drivers - where you can use your 'favorite' xslt engine. To me i think one issue with 4suite - is it is a REAL suite - i.e. its a LOT of junk we dont use" , although it did have support for "a Python implementation of the ODMG 3.0 object database standard" which Daniel Baumann (chillywilly) had been championing. Later, Nick reported "I appear to have successfully created pysablot packages" , in versions for both python 2.1 and 2.2, which were now available (http://nick.industrialmeats.com/proj/gnue/pysablot/) for testing. There was much gratitude from various people.

The next day (http://www.gnuenterprise.org/irc-logs/gnue-public.log.07Apr2002) , Jason declared that Nick "rocks because he created a pysablot deb for us :)" He had done some testing and "it appears to" work. Nick said "if you guys take on maintainership and clarify the license I'll upload itto debian" . It would also need "some documentation - there isn't so much as a readme in the tarball" . Jason suggested "who knows by the time we're done we might end up "adopting" a lot of small python projects" , as a "GNUe Python Orphanage" . Daniel suggested "GNUe Rehabilitation Center" .

5. Patching and profiling GNUe Forms

4�Apr�2002�Archive Link: "[IRC] 05 Apr 2002"

Topics: Forms

People: James Thompson,�Harald Meyer,�Daniel Baumann

James Thompson (jamest) said that "one issue with the win32 patch on multilines that jcater pointed out in his (and my environments) we have both win32 and unix clients - if a win32 client saves cr/lf to the db and a unix client reads it and modifies it then we have a problem :)" Harald Meyer (Harald1) said "one just has to make a workaround, when things get saved into the database" . He "might take a look, but I'm not sure if I understand enough of forms. How is the transfer from GFObjects into the database done?" . James said "i'll try and find some time over the next few days as this is going to be a pain in the rear - GFObjects never really hold anything - they pass it all to the data system - even forms not connected to a database create dummy datasources to store their info - so the edits would probably need to go into common" . He explained that the FieldDisplayHandler was "new - while an edit is taking place the value of the field is passing into the handler - once the edit is completed then value is passed back to the entry which store it in the datasource" . He had "removed internal event name translation into lower case today and make a cool error - where the event handler events were not being seen - so you go to a field , try and edit it (it'd go blank), then the old value would return on exit :)" It hadn't taken very long to spot this error!

He added "my test case was calling lower something like 11700 times - removing it from the event system saved a few calls - however" by writing the 'force to lower-case' as an exception rather than the norm "in one function dropped that to about 1200 - profiling is good :)" . Daniel Baumann (chillywilly) asked whether James had used "gprof" for profiling. James said "no - it's built into common's GBaseApp - forms, designer, reports, nav all based upon it" and accepted a --profile option flag from the command line. He explained "python has an internal profiler module" . He didn't think it used gprof, as "the profiler is written in python - and uses the way python works to hook into it - read the module manual profile section" . Daniel wasn't so sure.

6. GNUe Presentations

4�Apr�2002�Archive Link: "[IRC] 05 Apr 2002"

Topics: Why GNUe?

People: Derek Neighbors,�Arturas Kriukovas

Continuing the theme of Issue�#21, Section�#2� (14�Mar�2002:�GNUe demo at Phoenix Linux User Group) , Derek Neighbors (dneighbo) reported "i think get to do gnue presentation for local community college CS division next week i.e. one of those things where professors demand students attend kind of lectures" . Arturas Kriukovas (Arturas) said "not only Derek does this :) - on Friday i'll read a lecture in university about open source - linux & gnue main themes - but not for students - for professors :)"

7. Using CORBA and other RPCs with GNUe

4�Apr�2002�Archive Link: "[IRC] 05 Apr 2002"

Topics: Common, Application Server

People: Stefano Baronio,�Andrew Mitchell,�Derek Neighbors,�Calum Morrell

Stefano Baronio (steu) asked whether "geas is based on corba?" Andrew Mitchell (ajmitch) said that GNUe Application Server (GEAS) was "being rewritten anyway, so it can use corba, soap, xml-rpc, other funky system" . Derek Neighbors (dneighbo) confirmed "geas v1 uses corba yes" . However, the new version "will use gnue rpc - which is an rpc abstraction library - it will support XML-RPC, SOAP, CORBA probably right away - but its design is such that it can support any rpc that one wishes to write a driver for" . Stefano said he was "writing a generic system for distrib objects in corba, and I have found geas, but" wasn't sure if it could help him. Derek said "geas is not really distributed corba objects - i.e. our approach is to not lock ourselves into things - the idea is we use rpc as a transport only - not to actually deal with the objects" . This was true of even the first, CORBA-only, version of GEAS - "i.e. you defined objects via a class definition - that got loaded into GEAS but there was not individual IDL for every object" . Stefano said "my problem is not how to write a system like this, but how a medium developer can interoperate with it without knowledge of corba and c++. - for example, a VB or delphi developer" . Derek said "our view was that if you merely have an abstraction the developer really need not understand much of CORBA etc - as they have a lightweight CORBA api to deal with - instead of having to use CORBA to ineract with every object" . Stefano said he needed a multi-platform solution that was quick to develop on. Derek said "gnu enterpise doesnt have the goal of being a component company for vb/delphi (which it sounds like what you are looking for) - so i guess my short answer is that geas has little value in that context" . However, "there are plenty of folks that would discuss such things here as we discuss about anything" .

Later, Derek said "there is some misnomer that i love delphi - i think some things in delphi are great adn i think borland builds much better tools than say m$ - but i dont think gnue should be delphi" . Calum Morrell (drochaid) suggested that tools "like kylix" were better as free alternatives to Delphi as a generic rapid development tool.

8. GNUe as a free alternative to Microsoft Access - and more

5�Apr�2002�Archive Link: "[IRC] 06 Apr 2002"

Topics: Forms

People: Rob Adamson,�Peter Sullivan,�Nick Rusnov,�Derek Neighbors,�Dmitry Sorokin,�ra3vat

Rob Adamson (bobacus) said he had come "across gnue on the web a couple of weeks ago, and it looks like an exciting development - particularly if it can replace MS Access :)" , as previously discussed in Issue�#21, Section�#9� (18�Mar�2002:�GNUe as a free alternative to Microsoft Access - and more) . Peter Sullivan (psu) said "well, it's a bit more than that (or will be, anyway)" . However, "the beauty of GNUe Forms as an Access replacement is that you get to use *your* choice of "real" back end database - be it postgresql, SAP-DB, mysql, firebird or one of the non-free "famous names" DBs" . Rob said he was installing the Debian packages. Peter said "I'm not sure how "solid" the debs are at the moment" . Nick Rusnov (nickr) warned "the debs are being dropped from the distirbution I think because the latest ones are unbuildable" . Peter felt "It's a tad embaressing for a free software project (and a GNU project at that), but the "easiest" platform to install on at the moment is M$ Windows - just grab the setup.exe and go" . He suggested "I suspect you may be better installing from source on Debian GNU/Linux - I don't believe the dependancies are too awful to hack manually - and are reducing anyway ;-)" Nick agreed - "installing from CVS is very easy" . Later, Derek Neighbors (derek) agreed - "on debian get all dependencies from sid via apt and just get the cvs copy of gnue" .

Dmitry Sorokin (ra3vat) "wonders why so many peoples do not like access that much?" . Peter said that Microsoft Access was not a 'real' relational data base managment system, "but it gets (ab)used for fairly big d/b jobs because most people on Win platforms have it as part of Office - and hence it is "free" (as in beer) to them" . Rob said it "does horrible locking stuff with its mdb files, which seems to interact badly with samba, afaict - but it can be used with other back-ends via ODBC" .

9. Scalability of GNUe Application Server (GEAS)

6�Apr�2002�Archive Link: "[IRC] 07 Apr 2002"

Topics: Application Server

People: Andrew Mitchell,�Peter Sullivan

Andrew Mitchell (ajmitch) asked "should the different parts of GEAS be all running as separate threads in one python instance, or would you rather them be in separate instances, communicating via gnue-common libs?" . Peter Sullivan (psu) suggested "I would guess that GEAS should be as separated as possible as you might want to have it on multiple servers" . Andrew felt that "would make it more scalable, & robust, imho" . He noted "webware has an appserver, i think we shoudl look at that for how they do object caching, etc" . Peter said that "scalability is important" for GNUe, as "no matter what we say, some people will use it as a Quicken/SAGE/M$Money clone on a single PC [...] but at the other end you can replicate your GEAS servers as much as you like" . Andrew suggested "a node controller" to provide functionality such as "automagical failover, so that transactions aren't lost" .

10. Running applications from Navigator

8�Apr�2002 (1 post) Archive Link: "[Gnue-dev] Re: Running applications from Navigator (Kernel Cousin #23)"

Topics: Navigator

People: Stan Klein

Further to Issue�#23, Section�#16� (2�Apr�2002:�Running applications from Navigator) , Stan Klein said that, in so far as "Navigator is based on my business-process-oriented menu system proposal" , he had always intended "to include the ability to run general commands and to pass parameters to those cammands" for functionality such as:

  1. Viewing reports
  2. Getting program-specific help
  3. Accessing organizational policies related to specific business processes

11. Porting GNUe objects to XSD

7�Apr�2002�Archive Link: "[IRC] 08 Apr 2002"

Topics: Application Server, Common

People: Ian Stewart,�Jason Cater,�Daniel Baumann

Ian Stewart (Javaguy) "wanted to run past an idea re: porting GNUe business objects to XSD (http://www.w3.org/TR/xmlschema-0/) - XML Schema Definition" . Jason Cater (jcater) "thought xml schema was a replacement for dtd" . Ian said "XML Schema is being used to replace DTD's, but that's only one use - Basically, you have an XSD (schema definition), which describes a complex structure, and then you have an XSI (Schema Instance) for each instance of said structure" . Daniel Baumann (chillywilly) said "our schema language should be ODL as that's the purpose of it, to describe the schema of an ODMS (which include object application servers)" . Ian said that, although "ODL seems to make the most sense for internal definitions If you want GNUe to interface with .NET services, SOAP, etc, ODL is not going to do you much good..." . Daniel said "ah, but we have a layer to do remote communication protocols" . He added "we don't have DotGNU integration yet.... juets ways to call methods remotely" .

12. Working on GNUe Reports

7�Apr�2002�-�9�Apr�2002�Archive Link: "[IRC] 08 Apr 2002"

Topics: Reports

People: Derek Neighbors,�Jason Cater,�Nick Rusnov,�Reinhard M�ller

Derek Neighbors (derek) said he would use a sample GNUe Reports Definition (.gfd), and do "an html.xsl sheet as it should be easy - then i think i might make a gnumeric one - if i get VERY brave i will look at a formatted objects one which will segway into ps, pdf etc" . Jason Cater (jcater) "wants a text one, too" . He asked "how does everyone feel about standardizing on SVG for dynamically generated graphics (i.e., charts, etc)" . Derek said "i feel good about svg but i claim to know little :)"

Jason confirmed that the output parameter for GNUe Reports was "-p or --destination" . Derek thought this "seems like an odd choice :)" . Jason said "-d is taken for "debug"" but he was going to see if that could be changed. Nick Rusnov (nickr) suggested "why not -o" for output?

Derek asked what the <GNUe-report> and <GNUe-request> tags in the output were for. Jason said "you can ignore the GNUe-report-* stuff" , whilst "the GNUe-request are the options the report was run with - in case the raw xml was written to a file and later transformed" . He didn't really want to remove either of these "unless it really causes you a problem" - "nothing is permanent at this stage :) but my intent is to keep them" .

Later, Derek confirmed "basic transformatoin for title is done :)" He suggested "for now i want to be 'simple' so im going to ignore 'page breaks' in html - and instead insert <hr>" . Jason confirmed that section titles would be marked up as "<sectionTitle> foo </sectionTitle>" in the source XML. Derek was "just curious if we are avoiding <section title="something"> for a reason - i think its easier in xslt to do the way we are doing it :)" . He confirmed he had "subtitles and sectdions" working.

On formatting the tables, Derek thought "width, alignment etc should possibly not be in the [...] source file - this is where im torn - if we leave it out we are better off - BUT people will whine if we dont - i think the right way to do it is allow custom .xsl files to apply ontop of standard - but that another debate" . Jason said "I have ONLY one reason for doing that - I don't like it - BUT - this is a demo in which I need PS and Text demos - and at the moment, I can't figure out tables :) - so if you have a "width" we can fudge it for the moment. Now, if you know how to draw tables w/o that information - I'm game - long run, that's what I'd like - and remember, this is a proof of concept markup, not necessarily "official gnue"" . Derek suggested hard-coding a "<table width="100%">" in the HTML output for the moment. Nick Rusnov (nickr) suggested "use CSS, it makes size and formatting much easier" , but admitted this would only help with HTML output.

After some more work, Derek was "getting closer" with the HTML output, and posted "a sample (http://gnuenterprise.org/~dneighbo/monthly.html) " . Jason got all emotional - "my baby - she's becoming a girl" . Derek said "let me knock out subtotaling - and then i will look at respecting the attributes" . He did not "like how you ahve done subtotals :) it will make me work" . Jason asked "how would you like it? we can change" . Derek suggested XML source mark-up like "<subtotalRow level="2">" . Jason said "actually, I think I prefer that" .

Derek did some more research on "some xslt stuff - like attribute processing" . He noted "you can combine documents" using attribute values to contain filenames, and "we can do formulas in xslt if we so desire - hmm it can do sorting and the likes too" . He reported "well i have it respecting attributes :) - but its not pretty" , as the table width "of 100% makes it ugly" . Jason suggested removing this - "the table column widths are actual max character widths - which may or may not be relative in html output - but was mainly for text/ps" .

Nick reformatted the output using a Cascading Style Sheet (CSS) to show how this would work. Derek "looked at the version you made but struggling to implement it" , but noted "if i remove that text="css/text" works fine - prolly will choke the snot out of old browsers though" . Nick said the stylesheet would normally be a seperate file. Derek suggested "if you want want to pretty this up some" , he might be able to get "the xslt sheet" to apply the formatting rather than have to use CSS. Nick noted "you are getting logical information about the rows aren't you? [...] if you're getting that information, you can put it in class names - if its in class names you can mark up based on it" . Jason agreed, "as I don't wanna put css in" the GNUe Report Definition (.grd) format.

After midnight (http://www.gnuenterprise.org/irc-logs/gnue-public.log.09Apr2002) , Nick suggested XML tagging like "class="total"" . However, "it doesn't matter what you call it on tehe grd side, as long as we can guess logical information about the data and convert it to CSS classes" . For number formatting, "what you should do is do like spreadsheets do it - and say 'this item should be numeric' etc - and then we can say at the HTML level 'numeric classed things are right aligned'" Jason said he had been planning to "have something like <field type="date" format="y-m-d h:m:s.i">2002-04-01 00:00:00.00</field> - and the XSLT could take it as provided - or if the locale preferred, could convert the format" .

Derek said "technically it might be preferred to make xslt do formatting" . Nick agreed - "I say remove all physical formatting from grd - only provide logical information for the formatting side to use as hints" . But Derek did not "want to lock us into xslt - so i woudl rather put the tag in grd and have our engine do it - though it think when we get cooking we probably wont have anything but xslt" . Jason said "fwiw, I don't want to put TOOO much processing off into the XSLT - as it isn't exactly known for its speed - i.e., transforms is one thing, sorting and format masks are another" . Nick thought "definitely do sorting and char level formatting on the grd side. but leave alignments and stuff up to the xslt based on type/class hints" . Jason said "there's several instances where I've needed right-aligned text- or left-aligned numbers - however, I do agree that the "default" could vary based on type" .

Jason went to bed, and Derek and Nick discussed using italics or sans-serif fonts, making more changes as they went. As more people joined the channel, Derek gave a quick summary of progress to date - "the report is certainly lame but its really living and breathing i.e. it actually gets its database from a database - you have an .grd which is xml that is like a form to define what db and such to connect too - then it goes out gets the data and makes an outputted xml file - and we do an xslt translation to turn it into html" . He felt it was "a pretty solid start" .

Derek and Nick continued to adjust the format to see what looked best. Derek explained "subtotalling is not fully functional" yet - "jcater will probably have them workign tomorrow" . He started to experiment with outputting newlines and tabs for plain ASCII output, and posted "the start of a text version" before going to bed - "i just think jcater needs ascii output for some dot matrix forms at work - so wanted to encourage that it might be doable pretty quick" , although he admitted "it was easier to try txt as a second transformation" than formatted objects for Postscript or PDF.

After some sleep, Derek started on the Gnumeric format as well. He asked "i thought that gnumeric saved in xml? [...] but default is some binary" . Reinhard M�ller (reinhard) said that the default was just "compressed xml - via gzip" . Derek decompressed it and exclaimed "man is it 'verbose' - will have to do for loops with variables in xslt get it to output" .

The next day (http://www.gnuenterprise.org/irc-logs/gnue-public.log.10Apr2002) , Jason asked about progress on the text XSL. Derek said he "was going to write some 'functions' so col width="" is respected - and would justify off that" . Jason asked "can I send you a sample of what I'd like to see?" . Derek asked "what do you want for 'pages' in the .txt version - will a 'form feed' char work?" . Jason said that "will work" . Jason said "if you can get the text.xsl up to speed then that'll free me up to make the datasources respect those passed parameters - this is gonna be one helluva 0.0.1 release" . Derek said "we really need all the items released - forms, designer, reports etc - its been a while on all of them" . Jason said "hopefully this weekend or next week" .

Derek said that one of the "downsides to GNUe - every time you need info you are your own best reference (according to google)" . Jason agreed - "no kidding - I searched for xslt postscript yesterday - and guess what was at the top spot? gnue reports :) - which blew me away" .

Jason posted a sample text (http://www.gnuenterprise.org/~jcater/foobulations.txt) report, and asked "btw, is there any way in XSLT to store arrays?" . Derek said "um not really - its morelike a functional language than a procedurual one" . Jason recognised that, but "was wondering if it'd be possible to remember the "width" settings for the <col>" headings. He "only had those in the <heading> section - it seems SO wasteful to have to repeat those for each row - but if that's all we can do, then that's all we can do :)" . Derek said "you can have it persist" .

13. i18n and translating error messages

9�Apr�2002�Archive Link: "[IRC] 10 Apr 2002"

Topics: Common

People: Arturas Kriukovas,�Jason Cater,�Derek Neighbors,�James Thompson,�Nick Rusnov

Arturas Kriukovas (Arturas) suggested "maybe it would be better to assign for i18n one directory - put there translations (maybe in subdirectories), some kind of readme what i18n is, in what ways it's going to be realised and etc.) - and that everyone could put his ideas/corrections there" . Jason Cater (jcater) suggested the translations could go into "/usr/local/gnue/etc/lang/index.html" - or possibly "etc/locale" , to allow for localisations like time format as well - "and maybe have subdirectories under that for each supported locale" .

Further to Issue�#23, Section�#1� (27�Mar�2002:�i18n support for GNUe Forms) , Arturas reported "some good news from i18n world - i & Dmitry (ok, it was Dmitry on the whole :) managed to get i18n (=> Russian) symbols in wxForm under win - source still looks awfully, but it works" .

Derek Neighbors (dneighbo) felt that "documentation of what we are doing for all of i18n probably belongs in docbook/ - with proposals and other docs" . Derek felt there probably wouldn't be any tool-specific i18n files, so Jason suggested "to continue w/our current model, they should all go into common/etc" . This might not be ideal, but they could "worry about cleaning up when we clean it ALL up :)" Derek said "let Arturas check it in so we can blame him later when people complain ;)" . Arturas didn't like the sound of that, but Jason reassured him "don't worry.. we do that to derek ALL the time :)" .

Arturas asked "do we transalate error, debug messages?" Jason thought error messages needed translating, but "I wouldn't worry about debug messages at this point" . James Thompson (jamest) thought "i think the msg_blah stuff in gnue.conf needs to come out of there" . Arturas agreed, as then "there is no need to translate gnue.conf :)" Nick Rusnov (nickr) asked "does python support gettext?" . Jason said it did, but "unfortunately, we have to be careful in implementing that - as that adds a lot of overhead" as it "causes a python method call" . James asked whether they could be cached once on startup and "then the apps only reference the var that holds the output of the original _() call" .

Derek suggested that "if its a message that is useful to teh END USER it shoudl be translated - if its more a 'developer' message - it should be in english only - as we already are assuming all 'developers' will speak english at least enough to code in english :)" . He said "the problem i think lies in that our lower level debug statements - some of the stuff is actuall 'user messages' - but not all of it" . Arturas agreed - "all other strings (that end user can see) will be translated" .

14. Bayonne (GNU Comm) version 0.8 release

9�Apr�2002�Archive Link: "[IRC] 10 Apr 2002"

Topics: Bayonne

People: David Sugar,�Derek Neighbors

David Sugar (dyfet) said there would be another release of Bayonne this week - "It's the big 0.8 :)" . He confirmed that Bayonne contained some regression tests which could be used to check out problems. Later, Derek Neighbors (dneighbo) said "if you could post a story to main site i will publish it - i dont know what we had worked out re: downloads and such - i think we have a mirror of sorts setup" . David said "we may have 1.0 out by June" .

15. Jabber and Mozilla with GNUe

9�Apr�2002�Archive Link: "[IRC] 10 Apr 2002"

Topics: Forms, Common

People: Justin Kirby,�Nick Rusnov,�Daniel Baumann,�Jason Cater,�Derek Neighbors

Justin Kirby (return_this) said "I found GNUe yesterday and it peaked my interest" . He noted GNUe was written in python, and said "I am thinking about mozilla and jabber (my main points of experience) - does it make sense to integrate GNUe into these platforms? or is it too divergant?" .

Nick Rusnov (nickr) said "Currently most of it is in python" but "you already know python, basically, if you know C++" . He explained that, as python was interpreted rather than compiled, it "has a library management system similar to perl's" . But "for most things like simple scripts its just write and run" . Justin said "maybe perl has tainted my outlook... but scripts and I do not get along at all" - he thought this might cause stability problems for "something that will grow as large as GNUe will..." . Nick said "I see where perl would give you a bad impression of interpreted lanuages - Python is by nature object oriented - its not like perl umm package wise" .

Later, he explained "python goes like this "you use whitespace for what!?" then "okay, its not so bad" then "Wow, this is easy" then "I love python!"" . Justin said he had "just hit that "You use whitespace for what?!?!?!" phase" . Nick said "yea, thats the most painful phase of all - a lot of people get stuck there" . Justin said "also the lack of typesafety is troublesome too" . Daniel Baumann (chillywilly) said "that's what exceptions are for ;)" . Justin asked "are braces optional?" . Nick said "you don't use braces at all" . Daniel said "python seems very natural to me even coming from c/c++ background - I don't miss the braces" .

Earlier, Justin said he was looking at GNUe for "several things: a POS, Accounting for bother Retail and Service oriented biz - I am looking for a base to build off of for these types of apps... GNUe seems almost perfect" . Jason Cater (jcater) asked "how are you envisioning either mozilla or jabber working with business software?" . Nick said "I could see jabber for a distributed sort of business object network - you could use mozilla with a web-based forms client - but I don' see why it'dhave to be specificly mozilla" . Justin said he would use "mozilla for the XUL and XPCOM..." Nick wasn't sure about that, "although theres nothing stopping you from writing a XUL forms client" . Jason said that the FAQ (http://www.gnu.org/projects/gnue/faq.html) explained why GNUe was not keen on Java-style technology.

Later, Derek Neighbors (derek) said he saw GNUe using jabber "in two ways - a. as an option for asyncronous messaging in gnurpc (i.e. just one more rpc choice) - b. as a integral part of the groupware/workflow components" . On Mozilla, "i see someone could make an 'alternative' forms client with XUL model and java script" if they wanted, but he didn't "think its what core gnue developers should be focusing on" .

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.