Re: libstdc++ configure time

["Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>]

 > On Sun, Mar 18, 2001 at 08:04:14PM -0500, David Edelsohn wrote:
 > >       Complaining about default shells on various platforms does not
 > > really address the problem.  The slow configure problem has been
 > > reported since last Spring/Summer.  The other issue that was
 > > discovered is the repeated save and restores of
 > > AC_LANG_SAVE/AC_LANG_RESTORE in inner feture test loops.
 > 
 > That's already been addressed; we just test for lots fewer things now,
 > and thereby do fewer loops.  That's the only issue with slowness
 > that's appeared until now; nobody's complained since those changes
 > were made.
 > 
 > I've tried converting the entire script to use C++ by default (since
 > that's the environment under which all this will operate anyhow), but
 > got lots of weird failures in otherwise-good tests.  And again, the
 > part that causes slowdowns is where we're testing for correctness
 > during C linkage.  We have to change to C for that.
 > Phil


Have you thought about my proposal which allows you to do just one
AC_LANG_SAVE/AC_LANG_RESTORE no matter how many functions you test?

http://gcc.gnu.org/ml/gcc/2001-01/msg00971.html
http://gcc.gnu.org/ml/gcc/2001-01/msg01324.html

		--Kaveh
--
Kaveh R. Ghazi			Engagement Manager / Project Services
ghazi@caip.rutgers.edu		Qwest Internet Solutions