[Ffmpeg-devel] One fascinating read

Rich Felker dalias
Thu Apr 27 04:32:31 CEST 2006

On Wed, Apr 26, 2006 at 06:27:31PM -0700, Roman Shaposhnick wrote:
>   Not quite: the biggest impression I've got was that more and more
>   engineering projects in FSF are being hijacked by the activists

Hmm, what do you mean? My view is that the projects were hijacked
_from_ the activists (Stallman and so forth..) by people with business
agendas that conflict with the FSF's goals. GNU was a political
project to begin with. Saying it was "hijacked by activists" makes no
sense IMO..

>   of the worst sort. Come on, I would understand his point had 
>   glib been a pinnacle of clean code to start with, but he has no
>   right to blame others when it takes me 32 lines of code and 4 different
>   macros to just declare a function inside a glibc. glibc is scary

To some extent, I understand his point. I agree that it's
irresponsible of free software developers to make their code immensely
bloated, unreadable, and unmaintainable for the sake of compatibility
with broken proprietary software. But Drepper's analogies about the
majority and minority are fucking scary. I could very easily cause
Godwin's law to be invoked at this point but I think I'll refrain from

>   and rated "R" not because it has to support 1001+ platforms but
>   because the guy clearly doesn't care about writing clean code.

Precisely. Actually from my experience, much much more of the bloat in
glibc comes from the GNU extensions than from supporting multiple
systems. Many of these extensions were simple and easy to add when
they were first added, but then grew into horrible beasts over time
due to interaction with other extensions or with new requirements
(such as multibyte characters and locale) -- the whole "exponential"
complexity thing Drepper described, although in a different context,
and I somehow question where his "exponential" figure came from since
it sounds polynomial to me...

Perhaps someone should point Drepper at a clean libc implementation:


More information about the ffmpeg-devel mailing list