Re: OGo/GNUstep cooperation Re: Re[2]: Frameworks integration

[Helge Hess]

On 02.03.2004, at 01:11, Manuel Guesdon wrote:
> Note: it's probably better to keep the whole discussion here, in 
> discuss-gnustep

Yes.

> That said, my position is still the same than in december 1998/january 
> 1999, before starting GNUstepWeb:
> I think it's better to concentrate work on a single WO compatible 
> library. I've asked Helge about releasing their WO compatible library. 
> The answer was no and I don't want to discuss his (and mdlink ?) 
> reasons.

Yes, it was a company decision which isn't worth to be discussed. I 
hope you remember that I explained to you various things on how WO 
worked as well as contributed major parts of your initial 
implementation. Do you still remember who wrote NGAntlr initially? ;-)

> I can't find contribution from Helge/kyrix in GNUstep base, db, 
> extensions, gsweb ... ChangeLog but, anyway,...

Thats unfortunate. You should search better and check your personal 
mailing list archives.
MDlink, SKYRIX as well as me have a looong record of contributions.

> During 2003, Opengroupware appears as (L)GPLed. Great.

Notably skyrix-xml,core,sope where LGPLed long before OGo, somewhere in 
2001/2002. And MOF2 (more or less the predecessor of skyrix-core), was 
in 1999.

> Well, a strange way to start a cooperation as Helge don't know too 
> much about gsweb. But OK, may be a bad wording choice.

Not really. Why do you claim that I don't know about gsweb? I'm on the 
gswhackers list as you realized and I'm indeed tracking GNUstep 
development for years?

> Funny note about testing and usability, we have a GNUstepWeb have 
> application which is used by more than 40000 people each month, 
> serving more than 200 000 pages each month for near a year.

Yes, you actually have an application, great :-) I have no doubts that 
gstep-web does just fine for your specific case, so do you developer 
gstep-web just for that customer?
NGObjWeb has thousand of installed servers, not just one. And it had 
seven or more years to mature.

> I think Dave have similar exemple.

Dave?

> After reading last messages, I worry about the future of GNUstep 
> project as Helge see it. After years of proprietary development, he 
> sudenly suggest to share things.

You are really spreading FUD here. Just take a look in:

  ftp://ftp.gnustep.org/pub/gnustep/contrib/

This contains stuff like MOF2 which we shared in 1999. We paid Ovidiu 
for adding GC support to GDL and libFoundation before. As well as 
supporting Nicola on JIGS. Etc.
Get to the ground and open your eyes.

> But I don't think it's reasonnable
> to say "hey just drop this because we have a better one" without even 
> knowing exactly what we talking about.

I know what I'm talking about. While I'm watching gnustep-web the whole 
time, you didn't take a single look at SOPE, yet you are claiming that 
I do not know what I'm talking about.
Quite weird.

> seems inapropriate to me as Helge know, base parser is just around 500 
> or 600 lines of code easyly testable (just check your 
> NGObjWeb/Templates/WOHTMLParser.m).

Its excellent that you actually start sharing, we are making progress 
here ;-) Yet I do know about all the bugs and tweaks in the parser 
until it reached the current state. And this is just one, little part 
of SOPE.

> Nobody  (including) me does the effort to write a raw html parser. I 
> didn't
> done it because personnaly I like XML parser even if it can be 
> improved.

In short: the goal of your project isn't WebObjects compatibility, it 
is some other goals you have (but didn't line out). This is different 
for NGObjWeb - it was always top priority to be as compatibile as 
possible with WO. Of course the whole compatibility issues started to 
dimish with WO/Java.

BTW: Notably SOPE has a new approach to XML templates which is quite 
interesting and might be worthwhile for you to discover ;-)

> BTW, Helge, do you know about .gswd inclusion, automatic inclusion of 
> images,... in templated mail, query path extensions,...
> which are present in gsweb ?

Yes.

> So, after reading last posts, I'd like to be sure about the direction 
> of projects. If the goal is only to give more mainteners to OOg/Skyrix 
> projects, I don't agree.

You make it sound like a one way direction, as if gnustep-web 
participants can't gain anything from SOPE.

> I also don't agree if the point is to move some base things outside of 
> GNUstep tree.

Thats not a point. And this needs to be discussed. I have the 
impression that GNUstep itself is not very much interested in 
gnustep-web but more in the AppKit thing (though I pushed people to 
change that several times!).

So while I'm open to any suggestions, it might be worth to fork a new 
project focusing on a WO clone. Whether this is named gnustep-web or 
SOPE or xxx - I don't care.

> I trust Adam, Richard, ... because they work on GNUstep for years. 
> There's also few points about organisation, copyrights and so on but 
> that' a next step.

Well, "..." isn't very specific ;-), but yes, I agree :-)

> To summarize, I'm open, I'm happy to see contributions, I'll be happy 
> to merge NGObjWeb and GSWeb but I don't want to see GNUstep projects 
> taking some out of control way and I just can't say to people using 
> gsweb or to my
> customers  "hey, we'll drop  this so you'll have to rewrite and test 
> part of your applications; Are you happy ?".

So you are actually more bound to your customers than to providing a 
viable framework. Which is understandable but certainly doesn't match a 
community project like GNUstep very well.
I'm more interested in working on a general purpose framework.

Helge
--
http://docs.opengroupware.org/Members/helge
OpenGroupware.org