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

Re: Bug#139945: ITP: prokyon3 -- a multithreaded MP3 manager and tag editor for Linux.

On Wed, Mar 27, 2002 at 06:23:18PM -0500, Marcus Brinkmann wrote:

>> Joining RMS's "what's in a name?" crusade isn't going to make most
>> programmers care about portability.  In fact, NOTHING is going to make
>> most programmers care about portability, except when it's portability to
>> their own platform.  As someone who takes pains to make his code as
>> portable as possible, I can attest to the fact that writing portable
>> code SUCKS.  It's a lot of work for very little gain.

> Most people in the GNU project would disagree with you.  GNU software is
> very famous for its portability.  Portability is especially important
> for postmodern programmers.  Beside improving the quality of your code,
> it will also make your program compatible with the _future_ versions of
> your operating system.  So even if you only care about one system, you
> should care a lot about portability.

As I said, I try to make my own code as portable as possible.  But 
writing code that just has to work on a GNU system (glibc, gcc, gnu ld) 
-- that's easy stuff.  Making sure your code compiles on 32-bit and
64-bit platforms, and is endian-clean, is forward-looking, common sense 
stuff that does make for better code.  I'm not talking about that.  I'm 
talking about the kind of nasty stuff that autoconf and libtool were 
designed to help with.  Things like Unix platforms that don't ship with 
an ANSI C compiler by default; like having to carry around private 
implementations of fifteen different standard functions, just in case 
the platform you're building on is one of the many whose libc sucks in 
comparison with glibc; like having to support system-specific ioctl's 
for four different kernels.  /That's/ the kind of portability that sucks 
to write, and it doesn't make for better code, either.  Now granted, 
porting from Linux+glibc+gcc to Hurd+glibc+gcc isn't the biggest jump, 
either; and GNU tools like autoconf have gone even further to make 
porting work as easy as possible.  But even so, I will never be 
surprised to see programmers who make no effort to port code to 
platforms they don't use.

> I don't know of any case in the GNU/Hurd project where we asked non-GNU
> projects (or even GNU projects) to support us without providing
> appropriate patches.  But I know about upstream developers who have
> taken the initiative and fixed portability problems without letting us
> work out the fix.  And I know only about one serious dispute about a
> portability patch.  I respect your opinion, but I am also happy that
> actual reality of porting free software to a new operating system is in
> general well received in the community.

And all of that is commendable, on the part of everyone involved.  What 
is not so commendable is the sort of nonsense that goes on here on the 
mailing list with Hurd supporters making a ruckus about not crediting 
GNU -- here of all places, among a group of people who probably 
appreciate better than any others in the world how much the GNU project 
has done -- and ... bemoaning? decrying? those programmers who don't 
care about portability, as if this was some sort of an offense.

That's a far cry from the "thank you for your contributions to Free 
Software" attitude that our community reportedly prizes so much.

Steve Langasek
postmodern programmer

Attachment: pgp8D4_4xSlLi.pgp
Description: PGP signature

Reply to: