On Tue, 2 Apr 2002 14:30:59 +0200 (CEST) Alexander Leidinger <[EMAIL PROTECTED]> wrote:
> On 2 Apr, Simon 'corecode' Schubert wrote: > > >> Do you have test cases where the removal of icc's includes and the use > >> of "-X -I/usr/include/g++ -I/usr/include" fails? > > > > yes. just as simple as that: > > > > #include <string> > > std::string test; > > > > it also fails after fixing the wchar_t thingy. > > the headers from the base system don't use namespaces correctly... > > "test.cc", line 4: error: name followed by "::" must be a class or namespace name > std::string test ="Hello World"; > ^ > > So we have to fix the base system headers in this regard and discard > Intels C++ headers? > > What about these errors? > ---snip--- > test.cc: > "/usr/include/g++/std/bastring.h", line 142: error: too few arguments for class >template "reverse_iterator" > typedef ::reverse_iterator<iterator> reverse_iterator; > ^ > > "/usr/include/g++/std/bastring.h", line 143: error: too few arguments for class >template "reverse_iterator" > typedef ::reverse_iterator<const_iterator> const_reverse_iterator; > ---snip--- all these are because the headers from the base system are not standards conformant. i don't think fixing these headers will be an easy task. tomorrow i'll check out if i can somehow patch the icc headers and/or continue to patch the headers coming from gcc 3.0.4 cheerz simon -- /"\ http://corecode.ath.cx/#donate \ / \ ASCII Ribbon Campaign / \ Against HTML Mail and News
msg33255/pgp00000.pgp
Description: PGP signature