> I'd be grateful if you detail a bit about this compiler's peculiarities.
> To date, xlf is still my favourite, perhaps Microsoft can beat him... :-)
> I don't work with M$ systems at all (lucky me!), so I have no idea.
df cannot do any form of useful preprocessing; this build system uses
gcc -traditional (without #line directives) to do it. Basically xlf is
on par there though, I had to do the same thing for it... End result:
the generated makefiles have an intermediate stage for these two
compilers.
df does not understand paths with forward slashes. VC++ does, which
makes it quite easy to use under cygwin/make. df takes anything
anywhere on the command line starting with a `/' to be an option.
Something like `df foo/bar.f' will result in a message in the spirit of
`bar.f: unknown option; do not know what do with foo'. The forward
slashes must be converted into backslashes before df gets to see them.
This is doable under cygwin + make, but nowhere pretty enough that I
would have wanted to do it under autoconf, but then again, maybe it
isn't necessary. (Under cygwin the paths must be translated in make to
windows paths anyway before compiling with VC++ or df, as no native
windows product knows about its mount system and symlinks. But this is
rarely necessary under autoconf.)
Oh and of course it has different function naming conventions, you'll
have to force it to stick underscores at the end of the names and to
lowercase them if those are the conventions you are used to in your
mixed-language programs. The magic keywords are `-names:lowercase',
`-assume:underscore'.
I seem to be using `-c -Tffile.f -Foout.obj' to compile. Can't remember
why anymore. Oh, could be that it wanted `.for' by default, not the
standard `.f'.
I don't think I ever tried compiling F90 with this compiler.
This all applies to df versions 5.0, 5.0a, 5.0b -- I think.
Hope this helps,
//lat
--
But in our enthusiasm, we could not resist a radical overhaul
of the system, in which all of its major weaknesses have been
exposed, analyzed, and replaced with new weaknesses.
--Bruce Leverett, "Register Allocation in Optimizing Compilers"
