ssr@taylor.Princeton.EDU (Steve S. Roy) (08/23/90)
Howdy netters, I just got at&t's cfront 2.1 and I'm working on porting it to a MIPS machine. (That's one of the machines directly from MIPS Inc, not a DEC or an SGI.) I'm using a Sun 3 for the host machine. I've followed the manual as closely and I can, and tried every permutation where there's an ambiguity, but it doesn't seem to work. I can get it to compile (in one combination) but it dies with an internal compiler error when it tries to compile the libraries. What I've done is to first get a version working on the Sun 3. This is a `make world' type operation. Very smooth. Then I configure the files in a clean copy of the source for a System V machine, use the sun to make the scratch files (with the magic of NFS) and compile them on the MIPS. It builds a cfront, but when I try to use it I get an internal compiler error. If someone out there could give me a hint about making this thing go on something besides a 3B2 or a SUN, I'd really appreciate it. Steve Roy ssr@acm.princeton.edu
amoss@huji.ac.il (Amos Shapira) (08/27/90)
liu.se> <10629@cadillac.CAD.MCC.COM> <2045@idunno.Princeton.EDU> Sender: news@shum.huji.ac.il Lines: 17 Apparently-To: post-usenet@ucbvax.berkeley.edu Hello, I did a dirty port of cfront about a year ago. The compiler error seemed to be that the error reporting routine accessed the stack directly, and not through varargs, what I did was to avoid calling it. And it looks O.K. now. Maybe I'll fix it in the future by changing it ot use varargs (I didn't find varargs on the system then). Right now I finished porting G++ to the Iris, the libs compile but some of them die during the test (i.e. the test finds melfunctionning of some of the classes). Maybe you whould like to try this. Hope this helps, Amos Shapira amoss@batata.huji.ac.il