This is the mail archive of the libc-alpha@sources.redhat.com mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Wish for 2002 ...


Thus spake Thomas Bushnell, BSG (tb@becket.net):
> > No.  That's _exactly_ the point.
> > They would _not_ run better, because they still would have to include
> > a copy of strl* for systems which don't have it in glibc.  So you would
> > bloat glibc even further without any benefit whatsoever, because the
> > programs still need to include a copy of the code.
> They would run better on a GNU/Linux system, because the
> implementation in glibc would be better.

This is ridiculous.
If you touch that function, I'm not going to trust it and the glibc
implementation and will still use my own version of it.

And why would someone edit it?  That could introduce bugs and the
version is copied verbatim from the OpenBSD libc.  Others have felt no
need to modify it, why would glibc?  I put the functions in my diet libc
without modification although I am normally interested in small code and
good performance.  The API is so broken that it makes the most obvious
optimization (stopping going through the source string when you hit the
limit) impossible.

Optimizing these functions is like beating a dead horse.

> > Look, libc is not a masturbation exercise.  It is there to satisfy a
> > standard called the Single Unix Specification.  Programs are asked to
> > expect exactly the features of the Single Unix Specification and none
> > more.  If they know of some libc that offers extended functionality,
> > they need to test for its presence before using it.
> No.  A thousand times no.  glibc is not purely there to satisfy some
> wanking standards board.

Yes it is.  It is there to make portable programs work on Linux.
It does that by providing the API for it.

All the GNU software would not work if glibc didn't provide the APIs.
That is in fact the major selling point of Linux, that it is POSIX
compliant.

Your "let's add functions of dubious value so unportable crap code of
equally dubious value starts working" argument fits my definition of
wanking much better than the standards body.  What were your submissions
to the standards bodies so far?  You did several, right?  Otherwise your
position of talking down on them would be even more ridiculous that it
already is.

I suggest you get a life and stop pestering glibc with your toy ideas.
They already rejected them, it's time to move on.  For example, you
could learn how to write portable and safe code using strcpy and strcat.

Felix


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]