RE: The future of GNUstep (was: Open URL in NSWorkspace)

[Vaisburd, Haim]

Hi,

Sorry for the delay in response - I was travelling this weekend.

From: Richard Frith-Macdonald [mailto:richard@tiptree.demon.co.uk] 

>> In my mind, people who focus on Mac OSX compatibility [...]
>> do not expect independent development.

> I think this just shows that your mind is a long way
> out of touch with reality ... as I've never met anyone who thinks like
that.

Bad for my mind, but good for the world! I'm only happy if that's the
case.

> It's perfectly possible to aim for compatibility (for portability
purposes)
> while deprecating the use of poorly designed APIs.

Yes, and your work proves that. But I think that this aim became all too
important,
in other words, the priorities went wrong (striclty in my humble
opinion, of course).
Please read on.

From: Gregory John Casamento

> You could be more specific...
>
> 1) What breakthroughs?

I admit my handwaving here ( I mean, no real content ). More on this
below.

> 2) Focused and light-weight how?

When I wrote "I do not care about Mac OSX compatibility, because I do
not expect
to port their programs. I expect it other way around. Nor do I care
about Windows.
I promote Unix" I did not write the "but" - but I do care that the
toolkit of
choise can do the work. For writing new programs I basically expect two
things:

1. Feature completeness. For instanse, if I can resize the image, it
should be possible
   to rotate it, too. If I can transform the picture, I should be able
to save the result.
   If I read JPEGs, I should be able to read EXIF information that comes
along.

2. That the library does not break the common conventions on the current
platform.
   For instance, X Windows with all its remote functionality is the
current Unix
   convention. If, say, Emacs is usable with X over ssh, I expect other
graphical
   programs to be similarly usable.

Right now we do not have (2) - no fast remote X which also has
arbitrary affine transform and transparency, and we do not have (1) in
the tiny
area that I touched myself, that is, photographic image manipulation.
>From the posts in this list I got an impression that there are important
functional gaps in other areas, too.

So "focused" for me is first of all the question of priorities:
what is most important. For a Unix programmer who works on his own ideas
having (1) and (2) is the most important, while Apple Cocoa
compatibility is
of almost no importance at all, as well as Windows port, for example.
A focused library would be usable for real work (complete, reliable,
fast)
for at least one platform and one backend.

Of course, it would be good to have every proclaimed feature - works on
every platform,
Cocoa compatible, switches backends dynamically, and all that right and
fast.

But the history of GNUStep project shows that this is very hard to do,
and by trying to do it simultaneously we'd have the project obsoleted
before it's finished.

By "lightweight" I meant that having Apple Cocoa compatibility
might put some extra stuff in GNUstep just for the sake of compatibility
and is not, strictly speaking, required. 

>From many discussions on this list I got an impression that every
attempt
of porting any serious Mac OSX application failed - either because of
Carbon
or because of some toolkits on the top of AppKit that we do not have.
If this is a wrong impression because problems are reported on the list
but
successful ports go silently - please accept my apology. But if this is
true
- what do we really gain from compatibility?

I can't but cite Fabien Vallon:

> GNUstep is still running after Cocoa and still introduce new bugs etc
...
> This is pure waste of times / effort.
>
> [...]
>
> Why can we/you simply stay at OpenStep with few improvements :
> Example: Theming, autolayout, remove Archiver by KeyedArchiver, extend
services  (?)
>
> That would be enough for 90% of the applications [new applications
--tv]

It's weird you need my boring text to pay attention to these simple
words.

The only thing I agree with you and disagree with Fabien is that Desktop
functionality should be separated from GNUstep.

About "breakthrough". Unfornunately, I do not seem to have revlutionary
ideas,
but here are assorted thoughts that most readers, I guess, will find
trivial:

1. I think the focus shifts to handheld devices that will accept the
tasks
   now performed on desktop computers. These devices can be quite
powerful,
   but they require different user interface. Here is a room for
invention.

2. I'm a big fan of X windows. I think the same of event simpler
programs but on
   Unix with X windows can be more attractive than flushy but local Mac
OSX ones.
   For this, of course, we should not abandon X windows like Apple did.

   On the other hand, here are some ideas about future X development:
   http://people.freedesktop.org/~jg/Papers/ols2005.pdf ,
   they include interaction between handheld devices and other
equipment.

3. Like Etoile people, I'm not satisfied with the current state of
desktop.
   To my mind, it lacks modularity - "one program does one thing right"
principle
   is lost. There might be a room for experimenting here.

Here it is - this is as consice as I can get while still answering the
question ;-)
Sorry it went too long.

--Tima