[Info-vax] VMS port to x86

[John Reagan]

"Michael Kraemer"  wrote in message news:jpooau$kdu$1 at solani.org...


>Wouldn't it be even easier to have the Bliss compiler
>(and all other strange stuff)
>recoded in C, so it would be just another piece of C software
>which has to be compiled for a VMS build?
>You could keep your beloved Bliss based utilities
>but they would no longer be showstoppers for portability.

Having a portable BLISS compiler (and Macro compiler) would be a good idea. 
Several folks who read this newsgroup have looked into it.   The BLISS docs 
are pretty good for the language description.

However, for some reason, the VMS customer base uses other languages like 
COBOL, Fortran, BASIC, Pascal, and even Ada.  Don't forget them.

And just recompiling GEM with vaporware "portable" BLISS compiler, gets you 
an x86-hosted compiler that can generate code for Alpha (or Itanium).

"JF Mezei" wrote...

>And since x86 doesn't have the baggage of IA64 with epic and instruction
> bundles etc, it might be easier to use either Bliss-Alpha or even
>Bliss-Vax (and give it 64 bit) to create something which can take bliss
>source and make it work on x86.

The BLISS front is common source for Alpha and Itanium.  There is no "epic & 
bundle" baggage in the frontend.  You're back to talking about what code 
generator to hook upto the frontend.  GEM knows (or at least knew at some 
point in the past) how to generate 32-bit x86 code that conforms to the 
Windows Calling Standard, generating Windows object file with Windows debug 
information.  That would be one possible starting point.  The "Mister 
Potatohead" aspect of GEM is that the OpenVMS Itanium ELF/DWARF support 
could be also used as a starting point.

I've never understood all the negativity against the BLISS code base.  Yes, 
you can't easily find good BLISS programmers, but from the C code I've seen 
coming from young engineers, it seems we can't find good C programmers 
either.