Using a DFA for insn-recog.c

[Richard Sandiford]

Richard Sandiford <rsandifo@nildram.co.uk> writes:
> Mark Mitchell <mark@codesourcery.com> writes:
>> Also, based on the preface to your email above, I'm guessing that this 
>> patch might be providing leverage for some other improvement?  Perhaps 
>> that this paves the way for a future byte-coded implementation that will 
>> deliver faster compile times?  Or, do you think that the text size 
>> improvements alone justify this patch?  (I'm not saying they do not; 
>> just asking.)
>
> No, you're right.

Er, sorry for the ambiguousness.  What I meant here, and what I meant
when I said "the reasons you give" later, was that the justifications
for the patch were:

  - smaller text sizes
  - fewer predicate tests and shorter paths through the generated functions
  - paves the way for a byte-coded DFA implementation

Richard