Re: project goal Re: Release schedule

[Philippe C . D . Robert]

Hi Helge,

a little late, but ...

On Sunday, April 6, 2003, at 10:33 PM, Helge Hess wrote:
> Philippe C.D.Robert wrote:
> > Well, you cannot deny that there has been made much progress wrt the 
> > AppKit part of GNUstep in the last 1, 2 years.
>
> It has made a lot of progress regarding completeness of just the 
> implementation of just the GUI library. There is still little testing 
> by application level developers and even less testing by application 
> users.

That's true unfortunately. Personally I notice that I get almost no 
feedback wrt ProjectCenter (nor any other app I wrote). Now either they 
are perfect (...) or nobody uses them really, I guess it is the 
latter... :-| The question now is how can we change that and/or who can 
help us with this?

> The initial implementation time of some piece of software is only a 
> very small fraction of the time required to build a working system.

I know, but to do 100% we need apps and (more, new) users who test them.

> > > CF adds compatibility to MacOSX. Several things (eg XML and full 
> > > HTTP) are only available in CF but not in Cocoa but used in Cocoa 
> > > applications (because it's useful base functionality).
> > It does, but this is not related to Cocoa, so if we target Cocoa why 
> > should we adopt CF? I don't think we want/can/should 'copy' Mac OS X. 
> > Unless of course it can easily be done or/and somebody is interested 
> > in spending time for writing such an interface...:-)
>
> By targetting Cocoa I see "targetting Cocoa developers". And Cocoa 
> developers often need to use (and will use) CF on MacOSX. Eg Nicola 
> used CFXML on MacOSX for doing XML parsing.
> If you want to make it easy to port Cocoa applications, a CF 
> implementation is helping a lot.

I agree, but this would mean to change the scope of GNUstep.

> But: I do not think, that this is the most important thing. It's 
> "just" significant.

Yup.

> > GNUstep tries for completion, otherwise we can stop working on it 
> > right now. One step at the time, even if it takes a while... :-)
>
> But it misses a project plan on
> a) what exactly to accomplish
> b) in what timeframe
> c) how and with which resources
> Right now GNUstep is going on for about 10? years and the "only" 
> stable output is a Foundation library and a Makefile package.

Yes, it would be useful to update the schedule by assigning concrete 
tasks to volunteers. If we cannot do that we cannot estimate a 
timeframe, if we can then we see if the assumed timeframe is reasonable 
or not.

> > I agree, moreover I believe this is the only way to convince people! 
> > It is the solution which matters, not the technology.
>
> So shouldn't the GNUstep mission statement focus on solutions instead 
> of the technologies to accomplish them ? And line out what exactly are 
> the solutions' advantages against the existing solutions ?

I guess the fundamental issue here is that some of us see GNUstep as a 
cross platform API while others see it as some kind of successor to 
NEXTSTEP/OPENSTEP, thus it is not easy to address this with one 
"strategy". Radically speaking I believe either we try to create an 
entire system based on GNUstep or GNUstep ends up being used by very 
few developers as some kind of alternative to Apple, probably for 
GNUstepWeb or other non-GUI stuff. I do not think GNUstep will ever be 
able to catch up with Gnome or KDE as pure X11 desktop solution ever. 
Hopefully somebody can proof me wrong...

> > Discussion is needed, but talking alone does not solve the problem 
> > ...:-)
> Neither is just hacking into the dark ;-)
>
> IMHO GNUstep really needs some plan on what to do in what timeframe. 
> If we still talk about OpenStep compatibility in 2010, it's really to 
> late (IMHO it's already to late because other environments are well 
> established and are getting better any year).

It is IMHO too late for GNUstep to compete with KDE or Gnome, but it is 
(hopefully) not too late for a GNUSTEP/linux or a GNUSTEP/FreeBSD or 
even a GNUSTEP/Hurd (kind of the real successor to OPENSTEP/Mach) :-)

-Phil
--
Philippe C.D. Robert
http://www.nice.ch/~phip