I brushed off the portable fasdumper I wrote a few years ago and added
some tests, in the cross-fasdump-v2 branch.
It seems to work for cross-compiling on amd64 for amd64, at least, and
in principle it should work for any other instantiation of a
cross-compiler including different word sizes and byte orders and even
tagging schemes if we taught it. I suspect this would also make
transitions like like the ()/#F split and the scode type removal a lot
easier.
For SVM, I think we should add four new fasl arches: svm1 {32,64}
{le,be}, eg. svm1-64le for what you would get on amd64 or typical
arm64, but svm1-32le for what you would get on powerpc32. We'd need
to teach ./configure --enable-native-code about these four options,
perhaps with a default from the host's characteristics. What do you
think? Reasonable?
_______________________________________________
MIT-Scheme-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/mit-scheme-devel
