Re: sim/mips patch: add support for more NEC VR targets

[Andrew Cagney <ac131313@redhat.com>]

> At Wed, 27 Nov 2002 03:10:50 +0000 (UTC), "Andrew Cagney" wrote:
> 
>> This is the way it has (ment to) been done for all MIPS ISA variants 
>> since igen replaced gencode.  You've proposed a change to that process 
>> so I'm [trying to] explain the rationale behind the current status-quo :-)
> 
> 
> Ahh.  OK, I misinterpreted what you were trying to say.
> 
> So, all of that having been said, do you have strong objections to
> proceeding in the "new way" as described here in previous messages?
> 
> (You have the right automatically reserved to tell me "I told you so"
> if it turns out to be impractical.  8-)

If the function is going to contain:

	if (TARGET_ARCH (SD) == ...)

then I don't think it should be in the .igen file (i.e., put it 
somewhere else as is done with the FP code (where I can't see it :-)). 
The .igen files have a mechanism for differentiating between 
architecures, so I don't think that file should be confused by using a 
second mechanism.  If igen ever gets finished (the C code is replaced by 
a language parser) then the simpler that language is the better.  Part 
of that simplicity is avoiding dependencies on SD / sim-main.h like the 
plague :-)

As for having to tag each individual entry in the .igen file with an 
explicit CPU.  Yes, that sux.  However, I also believe that it has 
significantly reduced the overall error rate (no more breaking one 
target by editing another) and that benefit vastly outweighs the short 
term pain.

Andrew