[MIT-Scheme-devel] Re: changes to primitive definitions

[Joe Marshall]

On Sat, Sep 5, 2009 at 3:14 PM, Taylor R Campbell<address@hidden> wrote:
> You changed DEFINE-PRIMITIVES and a number of primitive definitions in
> the run-time system to eta-reduce them.  I presume that this makes the
> system run faster on your C# interpreter,

Actually, I want to avoid making any ad hoc changes to the main code
for my C# version.  The change for set-string-maximum-length! was
pretty much the only thing I couldn't work around.

But I have a ton of instrumentation in my version and I have uncovered
a few bugs that affect all the versions.  One example is the out-of-line
call to CAAR in ASSQ.

There is a bug I am tracking down right now.  It seems that there a
huge number of calls to the '%record?' procedure.  This shouldn't
happen because %record? is integrable and it ought to have been
turned into primitive calls nearly everywhere.  I think there is a bug
in SF that is failing to integrate primitives when they are part of
an ACCESS expression.

I figure that bugs that affect all the versions
(not just performance hacks for mine) would be useful changes.

-- 
~jrm