Re: [open-cobol-list] Cobol + GTK

[Brian Tiffin]

On Fri, 18 Jan 2013 14:58:49 -0500, Patrick  
<address@hidden> wrote:

> > Yeah, that's my bad and I'm in the midst of correcting.
> >
> > ocgtk.h is an empty header.
> >
> > a simple  touch ocgtk.h  will suffice to get you rolling.
> >
> > Sorries,
> > Brian
> Hi Brian
>
> Not to worry! I am a big fan of yours !
>
> You have tons of great stuff on the net and your probably the main force  
> behind my learning right now.
>
> Actually I have lots of questions for you....
>
> I have been following Vala for a few years and it looks great. The thing  
> is the project is very non-assertive. I mean it will be 20 more years  
> before they reach 1.0 and they talk about being experimental on their  
> site. I might be looking at projects spanning 5-10 years. I don't want  
> to refactor later if Vala is dead. Any thoughts on this, do you think  
> it's okay for production code?

I push it, but with the caveat that I also like to program useless  
trivials in address@hidden    ;-)

The Vala angle made certain integrations (GtkWebKit originally) a lot  
easier to manage.  I became a fan, but again, I'm a fan of many  
programming environments that are eschewed by the mainstream Java Python  
types.  I'd put Vala/Genie in production, but I work with a team of PHP  
Python programmers, so I don't.  They balked at REBOL and Forth, so I  
stopped pushing my opinions on them at work.  I leave that to lunch times  
and Friday nights around the pool table now.

> I was also thinking about objective-C and D for the GTK side, any  
> thoughts on that? We could bring Cobol in the same way you described for  
> Vala no?

Kinda.  The beauty of Vala is that it uses the same C intermediate method  
for compiling as OpenCOBOL's model.  That means that all the object files  
and share libraries are instantly compatible with OpenCOBOL CALL (and the  
intermediate C sources are cobc compatible as well).

Given that, I'm a fan of Walter Bright, he's bright, but the D linker  
isn't even compatible with C++ as the name mangling is different.  Adding  
extern "C" { } isn't that hard a wrap job, but it mandates a wrapper every  
time.  Objective-C will be in the same boat, with slightly different (but  
very important) details.  There are motions afoot to alleviate the name  
mangling C to C++ ABI problems, with ghost externals and hidden classes,  
but I prefer the good old C ABI unless it's not really an available  
option.  The Falcon experiment was C to C++ and back again, and it works  
well, though there weren't that many entry points to worry about.

LLVM clang works the beauty too, with the current OpenCOBOL source tree  
nearly untouched (one default gcc -f argument had to be removed).  That  
opens the future to that ecosystem.  Although I was pleasantly surprised  
at how easy an LLVM clang build was, I'm still against virtual machines  
that attempt to protect computers from programmers.  As a programmer, I  
should be able to code to the chip as rightly or wrongly as I feel like.   
If I'm not capable of more rightlies than wronglies, then I really should  
look for other hobbies.  But, for now, still a fan of the LLVM team, as  
they are pulling off some awesome work.

> When we bring Cobol code in, it is defined as an external  
> function/method. Do you think it might be good to bring the cobol stuff  
> in, inside a class? via a I guess extern private foo ...

That will work, absolutely.  I'm always writing OpenCOBOL experiments from  
the point of view of OpenCOBOL as master and ALL others as subservient.   
One language to rule them all.  It's a goal of mine.  OpenCOBOL as the  
hub.  That doesn't mean that OpenCOBOL modules can't be treated as  
subservient to other main applications.  Just happens to be my particular  
fanboy modus operandi.  (Well, the ROOT/CINT experiment pretty much  
dictated that CINT was master and loaded OpenCOBOL generated C sources,  
but I couldn't resist the funky cool interactive ROOT graphs and simply  
pretended that OpenCOBOL was still master of its domain, the domain of  
high energy physics visualization in that particular case).

> I am just outside of Toronto, if your comfortable with it, could you  
> tell me roughly where you are?

Ottawa.  Cold today, but warmer that yesterday.  -20C base temp is hard on  
the eyeballs, add wind and even the dog didn't feel like going out last  
night.  ;-)

> Thanks for responding to my post-Patrick

Patrick, welcome.

Cheers,
Brian