Hoorah! I have managed to compile the "rae" branch of CVS.
Many, many thanks to Jean-Marc for his suggestions about
integrating lyxstring with xtl, to Asger for passing my
problems with xtl to the xtl mailing list and to Jose
Orlando Pereira the xtl maintainer for his suggestions
about egcs.
Incidentally, I did NOT need to #include LSting.h in
config.h
Finally: a question. Can someone tell me how to use cvs to
create a patch of the differences between my stuff and that
in the "rae" branch. Something to do with cvs rdiff, but
I'm damned if I can get it to work!
Angus
JMarc> However, I had another nasty idea about these string problems: What
JMarc> about faking STL string with lyxstring when for the STL headers which
JMarc> require the <string> header?
JMarc> I explain. LString.h could start like
JMarc> #ifndef LSTRING_H
JMarc> #define LSTRING_H
JMarc> #ifndef USE_INCLUDED_STRING
JMarc> #include <string>
JMarc> using std::string;
JMarc> #else
JMarc> #ifdef __STRING__
JMarc> #error The <string> header has been included before LString.h
JMarc> #else
JMarc> #define __STRING__
JMarc> #endif
JMarc> #include "support/lyxstring.h"
JMarc> typedef lyxstring string;
JMarc> #endif
JMarc> #endif
JMarc> This way, any time the real <string> header is included, it will not
JMarc> load due to the fact that __STRING__ is already defined (if other
JMarc> implementation use other defines, we could provide them). The downside
JMarc> is that LString.h should be included before any STL header (we could
JMarc> force that by including LString.h in config.h...).
