optimisation??
Hi all,
I've seen quite a couple of packages, and was not too happy when I see
the -O2 -g cflags, with which was compiled.
1. Why do you use the -g flag? A simple user will NEVER debug a binary.
But wants it to run fast and be small. Because of the -g flag gcc will be
unable to perform some optimisations, and this does not make sny sense
since all binaries will be stripped - undebuggable with -g too.
2. Why use just -O2? The egcs people have been working for years to make
the compiler better. For us (too). -O6 makes faster binaries.
3. Why don't you use the -fomit-frame-pointer flag? The reasons are the
same as in 1.
4. What is the reason behind chmod -x the shared libs? With it, the so-s
will be un-ldd-able -> dephelper have hard time to find out the
dependencies, and nobody will be able to find out, what shared libs is
needed by another. I do not see any reason to make this. I use chmod +x
{,/usr}/lib/* after every update, and my system works perfectly.
5. I think it would be better if the debian/rules had a new target rule,
named config. -> no need to rerun the configure script if whenever
somebody wants to rebuild a .deb more than once.
6. Why are static libs in the -dev packages? Nobody needs static
libraries. It is an archaic thing, and (except in very interesting
situations) are not needed. (Or should gone to some -static packages).
7. My opinion: the most correct solution would be if the debian/rules
could handle CFLAGS and CXXFLAGS environment variables even if the
standard compilation scripts did not handle them - but would use -O6
-fomit-frame-pointer as default.
Thank you, and sorry for my english :)
bye
MaXX
Reply to: