I understand that egcs produced better code than gcc for x86 at one time, but this was never true for all architectures (example: gcc and egcs for ARM were always identical). Since egcs has been merged back into gcc, more and more of these special cased spec files are gonna be causing headaches in the future as well. If egcs is on a system, it is very likely the only compiler on the system, and already owns the cc (and gcc) symlink. At the very least, the following would do the same in a much more portable manner: if [ -e `type egcs` ] ; then \ make CC=egcs ; \ else \ make fi This will work even if egcs is a symlink (to gcc).
Uggh...that should have been if [ -e `type -p egcs` ] ; then \ make CC=egcs ; \ else \ make ; \ fi
Fixed in 3.5.11-1