Re: More lies about GS Programming from Turkey

[toddpw@ugcs.caltech.edu (Todd P. Whitesel)]

nathan@visi.com (Nathan Mates) writes:

>(2) Netscrape source code has nearly 500 C++ source code files-- a language
> for which there is *no* compiler on the GS at all.

Correct. However, there used to be C++ -> C translators (it's how AT&T did
the first few reference implementations of C++). It is possible that the
code could be crammed through one of the last releases of Cfront and then
brought over. Cfront was damned expensive, however, and porting the libraries
would still be a real tough nut to crack, though, so your eventual conclusion
that this "ain't gonna happen" is about right.

There's another problem which is far worse than the language issue: the 32
bit issue. Nearly all unix source code which has been ported to ORCA/C has
been a near-nightmare. Frequently the source code makes the bald assumption
that sizeof(int)==sizeof(void *) which isn't true under ORCA/C unless you
use #pragma unix 1 and pray real hard. Even then, those programs happily
assume they can chew 100's of K of stack space (ORCA/C does not offer any
features to allow this, you have to change your source), and do lots of
other things that just kinda don't work in GS programs.

If anyone was serious about porting a browser to the GS, I would start by
picking up the torch on the ill-fated "lcc" cross-compiler project, or
hiring someone to do a GCC port. Even if it required an interpreter or
pseudo-interpreter, it would at least have a chance of succeeding, and
we could re-use it to do other stuff.

My day job these days has me working on GDB (the debugger that comes with
GCC), and you might find it interesting to note that someone has already
written a PD 65816 simulator and taught GDB how to debug programs inside
it. FTP gdb-4.17 from ftp.cygnus.com if you want to fool around with it.
However, I do not see any "w65" support in the latest GCC. Oh well.

Todd Whitesel
toddpw @ ugcs.caltech.edu