On Fri, 2 Jul 1999 08:11:03 -0400 (EDT), Russell Nelson wrote:
>Sorry, Russ, but autoconf is a lose (compared to Dan's methods --
>obviously it's much better than the old "edit the Makefile; edit
>config.h; make; iterate" way). It's just wrong, all the way.
>Something can solve a problem but still be the wrong solution.
A problem with Dan's system [for others] is that the tools are not
released . Thus, for modifying his programs, one has to either reinvent
his wheel (ezmlm for some reason comes with the stub files that list
dependencies so it could be done) or go to the "edit Makefile by hand"
method. Dan's portability stuff builds an order of magnitude faster
than autoconf stuff.
OT: You can sell libc-based applications. I don't know what rights you
have to sell programs that use e.g. the stralloc() string
implementation (which is excellent, IMHO). It would be nice if some of
these things were available/public domain (it is explicitly stated in
the cdb package that one can do whatever one wants with the c files, so
a considerable subset are there). Also, a libdjb would save a bit of
space now that many of use run qmail, ezmlm, tcpserver, cyclog, svc,
cdb, ...
As for buffers, Dan's library is just a little bit more low-level. He
chooses ceratain buffer sizes in his applications, but that's not a
requirement of his library. I would think that his choice for his
applications is better than what designers of libc do at a global
level.
-Sincerely, Fred
(Frederik Lindberg, Infectious Diseases, WashU, St. Louis, MO, USA)
