Re: signature like constructions

["Robert I. Eachus" <rieachus@attbi.com>]

Simon Wright wrote:

> At some point you need (a) a machine-independent solution to the
> problem, (b) a machine-dependent optimisation. Clearly it's good if
> the machine-independent solution is also efficient, but it seems to me
> like overkill to write 6 very clever lines in the hope that the
> compiler (and future revisions) will always emit one machine code
> instruction. Why not use an assembler insert?

The LALR tool was used by Honeywell to make compiler front ends.  If 
others in the compiler group made a change to the Multics PL/I compiler 
that would break it, it would be regarded as a serious bug that would 
prevent the new compiler version from being installed in the development 
hierarchy, let alone as part of a shipping compiler.

I guess it is just a different way of thinking when you are part of the 
compiler development team.  For example, any changes I make to my code 
that is part of GNAT will only pass the regression tests if it compiles 
and runs correctly on all supported systems.  Similarly, I can count on 
any code that works once and makes it into GNAT working always because 
it will be part of the regression test suite.  It is not a joke that the 
automated regression test  suite, which may take hours to run, is one of 
the most valuable tools a compiler group has.  In many cases it would 
cost less to recreate a compiler from scratch, given the regression 
suite, than to create a new set of regression tests and the tools to run 
them.

-- 
"As far as I'm concerned, war always means failure." -- Jacques Chirac, 
President of France
"As far as France is concerned, you're right." -- Rush Limbaugh