Re: Hi, I'm back

[Riccardo]

Hello,


On Wednesday, May 18, 2005, at 06:40 PM, David Lázaro Saz wrote:

> I'm back.  After five odd years away from GNUstep development I can  
> return contributing again.  My circumstances have changed a lot,  
> obviously, from the beginnings of 2000.  I'm older and all that.   I've 
> been a professional developer all this years, mainly in  
> commercial/proprietary code bases.
Nice! it is good to get some experienced people back here. You seem to 
have left a couple of months before I got seriously interested in 
gnustep. I am currently not involved in the development of gnustep 
itself, I develop FOSS applications with it. But lately I contribute to 
GAP (gnustep application project) and anyway development and use 
uncovers always questions, bugs, things to get improved and this is my 
contact with the developer world.


> I have personal goals for this porting effort that could be at odds  
> with the current design of the GNUstep GUI library and parts of  
> foundation so I'd like to discuss my ideas with the people that are  
> now actively in charge of the Windows parts of the code.  Drop me a  
> message and I'll tell you what are my impressions or contact me  
> through AIM at this e-mail address (my time zone is GMT+1, Europe/ 
> Madrid, mind you).
Well, this is a good place to discuss public things and then we always 
have #gnustep on irc.freenode.org which is a nice place to chat too, 
although lately many "core" developers are busy or absent.

> Adding a little bit of detail to the picture, the main goal is for  
> this app to be completely integrated with the Windows look and feel.   
> I've developed some tests libraries in a variety of languages: C, C+ +, 
> Ruby bindings, you name it.  The end result is that there are some  
> things that I didn't think that could be done that are, well, doable  
> and working on Windows.  Things like activating XP theme support from  
> a DLL dynamically that I haven't seen done on other GUI libraries.   
> That was developed some months ago.

Well, if you already use EO, you don't use a native widget system, but a 
very well done theme Although I initially disliked this approach I now 
appreciate it more and think this is the way to go. It won't be probably 
be too difficult to make a NT4 like theme and one that looks like that 
rubber-candy look of XP.

I'm quite new in the window-side of GNUstep, I don't use personally 
windows, but lately I found interest in gnustep on windows in different 
occasions and for different reasons. A good gnustep on windows would be 
a very fine thing.

AFAIK, the current route is to work on MiniGW and use a native backendo 
on GDI. This certainly is the besto route to go to achieve maximum 
performance and integration. It will also ease deployment of gnustep 
applications on windows.
EO on the other hand took originally a Cygwin approac. I think that 
could be a good way to pursue too, for two reasons:
- it could be easier to have it working, since cygwin is much more 
unix-like and the backend could for example still work on X11
- some applications will need quite some adjustments to work on minigw, 
while probably you could have more luck on Cygwin.

So cygwin could not be only a temporary relief, but it could remain in 
the future also as "intermediate porting step".

To be honest my first and only attempt to build gnustep on cygwin failed 
miserably during the compilation of -base due to linker problems. The 
library archive files contain multiple entries with the same name and 
this confuses the linker. This is probably a limitation of the 
underlying .dll system, but I am really guessing in the dark here, my 
windows knowledge is a boundary of 0.

Unfortunately I had no time to report my problems and to have a further 
look at them.

> Then I looked into how that could be integrated in our current apps.   
> Looked through the active cross platform, open source, toolkit  efforts 
> without success.  After everything has been examined and  reexamined I 
> think that the easiest path, taking into account my  experience, is to 
> get the GNUstep code base and bend it as needed to  accept this.  That 
> way I will get what I need and maybe other people  can put that to good 
> use, too.

Of course no one wil stop you to "bend" existing code to your needs, but 
maybe you may want to have a look at what already exists and what 
possibly the mainstream solutions are and will be, that could help you 
to reuse as much as possible of the existing and future work in 
"official" gnustep, but also to be able to commit back most of your work 
and help gnustep itself.

> My main desktop/personal OS is OS X now and I've got a machine  
> available that boots to NEXTSTEP 3.3 so if someone needs something  
> related to that (testing, some questions answered, screenshots, for  
> example) drop me a line too.

I too have as main box a 10.2 machine, but I still have a 10.1 machine 
from which I am typing. But maybe more interesting, I have a set of 
odder and less common systems running a variety of unices. Some of them 
already support GNUstep and I try to regularly test on them most of the 
gnustep applications and gnustep itself (solaris on sparc, *BSD on 
sparc, 68k, ppc, x86, linux on 68, ppc, HP-PArisc, for example). 
Unfortunately some of these machines are a bit old and aging but often 
still helpful to uncover a load of different problems.


Have fun,
   Riccardo