horstman@sjsumcs.sjsu.edu (Cay Horstmann) (02/03/90)
About a week ago, I asked about porting AT&T cfront 2.0 to the Sequent Symmetry. All I got was a bunch of "gee, when you find out how let me know" responses. Thanks, folks! Well, today we succeeded finally. We never did get the method of first in- stalling 2.0 on a Sun and using that to build 2.0 on the Sequent to work. It would always bomb with a ridiculous "array too large" when building munch (the array in question was a struct _iob[20].) But we were able to first bring up 1.2 and then 2.0. The main problem was that the Sequent C library is horribly deficient, and we had to add some string functions as well as memcpy. It turns out that once we had 2.0 running we had to use it to recompile memcpy once more and add it to libC.a, since it is needed by the stream stuff, with a properly mangled name. The other problem was shell scripts and make files. There is no test -x on the Sequent, change to test -f. The make install in the end won't work at all. Use make -n install >out and sh out. The CC script has a -L option for the loader, which the Sequent loader doesn't support. Gotta move libC.a into /usr/lib or /usr/local/lib and eliminate reference to the -L flag from the CC script. That is all I can remember now. Hope it is helpful to others facing the same ordeal. Cay