[MLton-devel] mlton/doc CHANGES,1.62,1.63

[Matthew Fluet]

> I really have to figure out how to work around the problem with the shared
> gmp library.  Several times now I have had programs that don't use any gmp
> and when I moved them around to older systems they would not run because they
> don't have gmp.  My work around was to run
> 	ar qc libgmp.a
> (which makes a libgmp.a which is `empty') and to then run MLton with
> 	-L`pwd`
> at the end.  This caused it to use the local libgmp.a instead of the shared
> one so that the program wouldn't look for the missing shared library.
> Not a disaster, but not good.

Is this still a problem (i.e., with the current source tree)?  I resolved
one problem with the native backend where the arguments passed to gcc for
the compilation of the C-stub (which includes allocating int inf
constants) didn't include enough optimization options so that the call to
the initialization of int inf was eliminated when there were no int-infs
used.  A more recent version of gcc seems to do the necessary dead-code
elimination without a -O1 option.



_______________________________________________
MLton-devel mailing list
MLton-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlton-devel