> Am I wrong, that the ANSI C++ library is on very few systems (namely
> sgi) available as a manually optimized and thoroughly debugged option?
AFAIK, yes.
> Using this means asking for complaints for bugs/slow performance. So it
> might be a good idea to provide (precompiled and tested/debugged)
> versions of the library for the different compilers, as well as
> performance reports for all those ...
>
> You should also report, which compilers are 'LyX proof' and test for
I think one should only use features supported by more than one
compiler, even if this means you can't use all of these fancy new
gimmicks in the standard.
LyX's code does not seem to be monolithic with respect to coding style.
I think it would be much wiser to improve the oldest bits first than to
sprinkle bleeding edge iterator techniques into pointer infested source
code from the 70's.
gcc 2.7 is pretty standard outside the linux world, it is *fairly*
modern, so *I* would say this make a good yardstick...
> frustrations, soon stabilize an industrial quality code base and enable
*grin* industrial quality? I always thought LyX could do much better.
Andre'
PS: You know that I always sound harsher than I mean it, but it's Friday
anyway ;-)
--
Andre' Poenitz, TU Chemnitz, Fakultaet fuer Mathematik
[EMAIL PROTECTED] ... +49 3727 58 1381