On 02/11/2007, Joerg Schilling <[EMAIL PROTECTED]> wrote: > [EMAIL PROTECTED] wrote: > > > > > > > My main issue with autoconf is that its input appear to be completely > > wrong. Rather than telling it "I want to use socket(), where do I find > > it" it depends on people to hard code all possible OS variations > > themselves. > > Well, this is only partly correct. > > Many autoconf additions from people that do blindly follow FSF documentation > work the way you claim. This does not make autoconf itself behave this way. > > If you look on how all "schily" tools use autoconf, you will find that it is > possible to use autoconf in a real abstract way for many caeses. > > There are however areas where this won't work. The way different OS store > sub-second time stamps in struct stat is such a case. > > If you like to have portable software without reinventing the wheel every > time, > I recommend you to look at my include files in .../include/schily/*.h > They provide reusable code and if there is a bug, you just change one > .../include/schily/*.h file to fix it. > > The autoconf documentation that tells you "if you like to use XXX copy these > 10 lines into your source" is of course bad.
Have you ever thought of bundling your whole build system into one toolset like autotools has? Offering that as an alternative on OpenSolaris might be a great step in attracting developers. -- Shawn Walker, Software and Systems Analyst http://binarycrusader.blogspot.com/ "We don't have enough parallel universes to allow all uses of all junction types--in the absence of quantum computing the combinatorics are not in our favor..." --Larry Wall _______________________________________________ opensolaris-discuss mailing list [email protected]
