On Sun, Feb 22, 2004 at 03:18:22PM -0600, David Phillips wrote: > Munir Nassar writes: > > i highly doubt i18n is the cause for codebloat. The demands for > > eyecandy on the other hand is much more likely to cause a system to > > crawl. > > glibc is bloated because it's a GNU project. But seriously, it does seem to > be quite bloated. > > Compile a "Hello world" program in C, link it statically then strip it. On > Debian with glibc it is 416k. On FreeBSD it is 46k. That is a huge > difference. With dietlibc it is 12k. So you say a hello world prog is 416K with existing glibc? Does it consume that much in memory? How can I find out? Top seems a bit coarse, /proc/pid/mem seems like, hmm, just a bunch of numbers... If everythings dynamically linked, I would hope the glibc code space is shared by all apps, that would make it less of an issue. Karl. > My guess is that the primary cause of this is glibc not being broken up well > into modules, so lots of unused stuff is being linked into the binary. > Sure, a few hundred kbytes doesn't seem like a lot, but it can make an > enormous difference for many applications. For example, a very small > forking web server can outperform Apache and even some non-blocking web > servers when it doesn't have a bloated libc. > > -- > David Phillips <david at acz.org> > http://david.acz.org/ > > > _______________________________________________ > TCLUG Mailing List - Minneapolis/St. Paul, Minnesota > http://www.mn-linux.org tclug-list at mn-linux.org > https://mailman.real-time.com/mailman/listinfo/tclug-list > _______________________________________________ TCLUG Mailing List - Minneapolis/St. Paul, Minnesota http://www.mn-linux.org tclug-list at mn-linux.org https://mailman.real-time.com/mailman/listinfo/tclug-list