[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: What about a non-free compiler? (Re: new port: debian-win32. when ?)



David Starner wrote:
> 
> 
> It seems you avoided this whole question. What could we do about bugs
> in the binary caused by the compiler?
> 

If the non-free compiler has such bugs, there's almost nothing to
do, you could call up the compiler vendor and beg. Perhaps it
works, but in most cases ther would be a barrier between the
user and the developers.

> 
> Why? gcc has one of the most open developement systems of any software
> out there, and is currently lead by Mark Mitchell of CodeSourcery, LLC.
> Or do you believe that Debian is run by Red Hat to divert people from
> Red Hat's total conquest of Linux?
> 

I'm only skeptical. But if you would ask me, I would think
that gcc might be getting a bit outdated. Don't get me wrong, I've
seen its code. I think it's perfectly well written, and it's
one of the best compilers, at least in portability. On the other
hand I don't believe in any development model blindfolded! Over
the years, I've made such a painful transition to C++ such that a big
portion of my skin still bears its marks. I've had code crashed
Solaris g++ & Linux g++ since 2.7.x, and it still doesn't give
me the feeling of a big change in the framework. I would definitely
know the thing if it were there. :) I mean some of the things
still seem to be put there ad hoc. I'm afraid that free software
sometimes approximates to a greedy search that converges on local
optima in the design space . :I

> > I know of
> > all the arguments about the rights and everything belonging to
> > FSF and that it is pure free software, but what can I do if it doesn't
> > satisfy my requirements?
> 
> Fork the packages. Pay someone to fix them for you (CodeSourcery will
> be happy to do this.) Go work on another free C++ compiler (Tendra is
> in Debian.)
> 

No offense but, I once tried some net sites that promote the
"Bazaar" model of software development, including CodeSourcery.
I encountered some slightly enjoyable projects including a gcc
improvement, but who does that for $60? And I don't have the
time and the mood to fork gcc, or work on someone else's compiler.
Yep, because a compiler can be very personal. I'd rather make
a facelift to my parser, add a lexical analysis tool, and pipe
to a working backend. But I have no idea how long it would take
to get the free standard library to run. It seems beyond my
means to jump on another huge standard right now. Last time
I tried to do that it took me more than a year, so I don't
think it's a nice idea to lose more hair.


> > On the other hand, if g++ is becoming a conformant ISO Standard
> > implementation,
> 
> It's working that way. The only major thing in 3.0 should be the
> export feature, which isn't terribly useful for free software.
> 

Stop there! Is that true? Is that feature available in the version
under development? God, but that wouldn't be useful without the
whole logic changing, I'd like to see export keyword making into
libraries. That's the thing I need because a *good* export
implementation would be the only way to go for a library
making use of compile time polymorphism. I'd also love the
"named return value optimization", and also more of the standard
lib ironed out.

And it isn't terribly useful for free software now. But if standard
C++ is coming in with the g++, it's coming along strong. I've
always thought that a high quality C++ implementation has the potential
to be the language of choice for systems programming.

Sincerely,

-- 
 ++++-+++-+++-++-++-++--+---+----+----- ---  --  -  - 
 +  Eray "eXa" Ozkural                   .      .   .  . . .
 +  CS, Bilkent University, Ankara             ^  .  o   .      .
 |  mail: erayo@cs.bilkent.edu.tr                .  ^  .   .


Reply to: