There are a few problems with the makefiles:
* The UNIX makefiles are building executables with
".exe" filename suffixes. Since the MS-WINDOWS
makefile is separate from the UNIX makefile, this can
be fixed with a search-and-replace.
* "gmake" should be the same as "gmake all".
* "gmake all" [...wait...] "gmake all" results in
things being built TWICE. You've defeated something
fundamental in make which is supposed to only
re-process changed files.
* When building the TNT stuff (as part of "gmake all",
it blows up because you forgot to include the math
library (-lm). Perhaps your system rolls the math
library into the main library, or perhaps you have
"-lm" implied by your environment. It worked for me to
just append this to the definition of CPP.
* Related to the above, I have found that "gcc"
doesn't always realize that it's supposed to be doing
C++. So it's better to use "g++" rather than "gcc" if
you are explicitly seeking the C++ compiler. (If, on
the other hand, "CPP" was meant to refer to the C
Pre-Processor, then never mind.)
Logged In: YES
user_id=620670
I really only use this on Windows. When I test the unix
version of the build, I do it on cygwin which does a lot of
stuff behind the scenes.
I haven't put any real development effort into this version for
quite a while. I'm working (or thinking about working) on a
completely different implementation based on Softwire.
Since I never had users before, I didn't need to keep
anything up to date. If you are interested in using calculus-
cpp, then contact me directly at Oliver@Canada.Com