On Mon, 1 Apr 2002 10:55:22 +0200 (CEST) Alexander Leidinger <[EMAIL PROTECTED]> wrote:
> On 31 Mär, Simon 'corecode' Schubert wrote: > > > that's right. but as terry said, it seems to be a bad idea to use some > > headers that don't fit the libs. > > i tried to patch the headers coming with 4.4-S and lang/gcc30 (because > > the one from the base system doesn't provide a proper c++ stl > > implementation) > > From a C++ POV this may be better, but don't you think we better fix the > base system instead? this would be nice, but gcc 2.95.3 (from 4.4 base system) can't cope with several things so that the header files are not yet that sophisticated. gcc 3.0.4 on the other hand ships with lots of more standardized headers. if i am informed correctly, 5.0-c uses gcc 3.0.4? i don't think we can use gcc 2.95.3 with the 3.0.4 headers (didn't try tho) > > my first tries for the patched include files can be found here: > > <http://corecode.ath.cx/~corecode/stuff/icc-patched-includes-0.1.tgz> > > We should either fix the base system, or ${IA32ROOT}/includes/. i'd really like to, but the headers from the base system are just not standard conformant enough. the headers coming with icc just don't fit our libs; they are specially crafted for use with redhat linux 7.1 :/ i already tried to fix these, but our implementation of the multibyte functions won't let this be an easy task (if it ever can be solved): eg. class ctype relies on these functions not being implemented and lots of other classes in turn depend on ctype... i'm not sure if this will be any better than fixing the gcc 3.0.4 headers, but lets see... cheerz simon -- /"\ http://corecode.ath.cx/ \ / \ ASCII Ribbon Campaign / \ Against HTML Mail and News
msg33239/pgp00000.pgp
Description: PGP signature
