Re: project goal Re: Release schedule

[Helge Hess]

Hi Nicolas,

sorry, I do not have the time to go into detail on each of your points, 
and frankly I do not think that we can reach a common state anyway. 
Nevertheless I try to point out some things ;-)

I think we can stop the thread as replies to my (every extreme) mail. 
The mail was intendend as an attempt to urge people to think about other 
problems than just "should we implement OpenStep or Cocoa", the further 
discussion should go between the active contributors to GNUstep which 
obviously need to find a new, clearer mission statement.

Nicolas Roard wrote:
> I disagree. A GNUstep-gui 100% stable would be an equivalent of Cocoa;
> do you
> really believe that Cocoa is just slightly better than KDE ?

I do not like to argue about that "would be" because GNUstep will never 
be a 100% equivalent to Cocoa, if you only have 90% of Cocoa it's only 
slightly better than KDE, yes.

BTW: you are asking for a comparison of different things, the 
environment and a UI library.
Cocoa/gnustep-gui is comparable to QT
MacOSX            is comparable to Linux+KDE

> KDE is great, yes. But on the other hand, we have Gorm and Objective-C,
> plus
> a very, very, good API. I don't think we will overthrown KDE or GNOME,
> but we could easily become
> the 3rd desktop. We have the technology; what we miss is marketing.

Yes, blame marketing - the way the project goes on for more than ten 
years can't be the reason ;-)

> And on the user level, LinuxSTEP is one of the most interessting project
> I could see -- it could be equally interessting if based on KDE libs ;-)
> but I think that
> a Desktop inspired from NeXT IS a better desktop than one inspired by
> Windows.

I do not disagree with that.

[some things deleted]
> I'm pretty sure that people will come aboard when the project will reach
> a more visible state and stability.

So we agree that GNUstep is not even stable for *developers*. But the 
real QA phase starts (the 90% of the work) if people actually start 
implementing more complex applications on top of it and then, if *users* 
try using them.
Sorry, but IMHO you clearly underestimate the time which is required for 
that.

> I agree we sorta miss a component system (so does Mac OS X). But I think
> you
> overestimate the importance of a component system such as in KDE.

I don't think I overestimate that. It's not mission criticial, but one 
of the several things missing that make a solution complete.

> What
> could be done *now* for us is to program some additionals palettes to 
Gorm.

First you need to understand why fixed positioning systems are not 
working very well on X11 (and why X11 systems are different to NeXT or 
Apple computers). You'll find out why any other window toolkit uses a 
box system.

> Anyway, a "real" (as in embed any component in your app) component
> system doesn't seems very difficult to do using ObjectiveC and
> DistributedObject.

Myth. Writing a lot of nice things is going to become trivial as soon as 
we have implemented all features. This is not going to happen.
Things actually *are* easier than in other frameworks, but *not* in 
factors (if you are twice as fast in ObjC, you are good).

> But I think the simple "palettes" road is enough for most uses.
> Embedded components is generally a bloat.

If we cannot embed Mozilla, we need to write a similiary good HTML 
viewer (KDE did with Konqueror). If we cannot embed a WebDAV storage, we 
need to write a similiary good one. If ...

> Configuration management, I don't understand ... we have the defaults 
base
> (which is good) and we have an application such as Preferences.app,
> modular,
> wich is exactly what we need.

And how do you configure the rest of the Linux system ? I'm talking 
about the "Windows Control Panel" which is required for a desktop 
environment.

> Package Management ??what's the relation
> with GNUstep ?

See above, you aim to replace KDE, not Qt. Just Qt provides little value.

> it's a distribution-related program. And, we have anyway a
> good (very good) thing with our system of app bundles -- anything 
embedded
> in the app bundle, multiple languages, even multiples architectures ...

This is not package management. APT or RPM are package management 
systems and required for distributing a desktop environment. A major 
feature of Ximian/GNOME is the RedCarpet system which allows you to 
install and update the complicated GNOME system in a simple way.

BTW: this is something which is really lacking in OSX.

> A multimedia framework would be good, I agree. But on that particular
> point,
> KDE or GNOME aren't very well mature either.

Did you ever use current GNOME and KDE applications ?

> Well, we could follow 2 paths : provide a backend-thing which will use
> the local set of widgets (windows, qt ...) ...

No, that wouldn't do anything good on the real problem: missing 
resources. Basing on top of Win,KDE,GNOME only works well if only 70-90% 
compatiblity to AppKit is acceptable. Otherwise it will turn into even 
more work.

> I think it could be done
> (see wxwindows). But that's not the path choosen actually; people have
> more or less agreed that GNUstep apps will be consistant *across* 
platforms
> and not *on* the platform. Personnally, I will prefer consistancy on the
> platform..
> Anyway, an average path could be done using themes.

Themes for emulation do not work. Proven by Mozilla, by Swing, by 
OPENSTEP/NT, by ... Themes almost always only provide the same look but 
almost never the same feel.
Themes are good for customizing a UI (the way themes are used in GNOME 
and partly in Mozilla), but not for "emulating" another one.

> You seriously miss the point here. What had changed is the *look* of the
> UI.
> But the UI itself isn't so great.

You are seriously missing the point of your project members. Richard as 
well as Philippe have pointed out that the UI is the (their) major focus.

> With GNUstep hopefully, we'll have
> great UI for
> our programs (at least compared to KDE/GNOME apps) if we follow the NeXT
> guideline for example.

See above, several people want an exact replacement of the NeXT UI.

> Many people don't see behind the look, and it's normal. And I agree that
> if we want the project a little more visibility, a theme support,
> or at least another theme than NeXT, is mandatory.

Themes do not work well at all with the fixed positioning AppKit 
technology but need a box system. Claim ;-)

> see http://www.roard.com/screenshots/screenshot_theme20.png for example
> for my current tests. It's far from beeing finished, but it's *possible*
> to add theme to GNUstep quite easily.

Well, I do not see how, but I'll take a look if it's ready ...

BTW: you are going to get sued for the presented image ! ;-)

> That's why I totally disagree. Of course the current gnustep-gui is not
> finished,
> but it's basically works -- I'm using GNUMail as my main Mailer since
> *last year* !

I tried to use GNUmail, but GNUmail didn't cover my needs, 
*unfortunatly* no replacement for Moz-Mail (but could be better than 
Sylpheed or KMail). GNUmail is a good example that the UI needs to be 
rewritten between OSX and Linux anyway.

[KDE, GNOME wrappers]
> Well that's alreday done. KDE and GNOME for Objective-C exists since a
> long time. No one use them.

Yes, none is complete nor developed in an active way ?! Those wrappers 
of course also need a significant amount of work ! But this will take an 
order of magnitude less time.

>> If that would be done, 5 people could finally start out writing
>> applications in Objective-C ! The model layer could be easily shared
>> between OSX and Linux applications.
> Arg, it will be a pain in the ass, seriously...

Decision between write something or write nothing ;-) We cannot find 
agreement on this if you insist that GNUstep GUI is stable for writing 
applications.

>> And we can write good GNOME applications in ObjC much faster than
>> other people in plain C.
> Oh yes. But personally, I really dislike the GNOME project :-P
> I think it's doomed. Perhaps even more than GNUstep ;-)

Well, I also think that KDE is much more streamlined, but GNOME is still 
much further than GNUstep. Eg (not to lower someones work !) Evolution 
is a much better mail client than GNUmail. This is not because the 
people are smarter or the technology is better, but because the 
technology is not *ten* times worse (only 50% or something) and 
significantly more resources are thrown at it.

> We could do that using GNUstep. In fact I think we could do that in
> providing
> small apps, *cooperating* with each others, and we could end with a 
better
> desktop experience than KDE/GNOME. That's my opinion ...

Could, would, if ... your reply is full of it and the points are years 
old. Four years ago both, KDE and GNOME were not as far as now, now it's 
almost impossible to become a third desktop environment even if you 
throw loads of resources at it.

All this somehow reminds me of the various attempts to regain the Amiga 
OS. Amiga OS was great ... at it's time, now it's hopelessly outdated.

> Yes, I really love gsweb, and we could do great things with it.

"Could". You need more resources to "do". If *all* GNUstep would commit 
their time to gsweb, it could become awesome in a year. If the work 
keeps being done by Manuel alone, it will still be good but won't become 
awesome.
BTW: a lot of the things which are nice about WebObjects are currently 
implemented in ASP.NET which is going to be implemented by the Mono 
project. And the Mono people are *very* focused and have resources.

> Well, as it's an opensource project, you can't force people to work on
> things ...

And I won't ;-)

> They need to be convainced :-)
> I agree with your ideas concerning Objective-C and GSWeb, I'm all for it,
> but that doesn't means I will stop working on gui apps ... ;-)

Yes, I know ;-) I know that gnustep-gui won't be stopped, because it's 
nice hobby-time hacking and even looking at an incomplete NeXTstep UI on 
top of Linux is plain fun.

But if you think that you can accomplish a third Linux desktop 
environment without good project management and project plan in 
sparetime, then sorry, you are unrealistic.

regards,
  Helge