Re: [Sbcl-devel] Re: [Sbcl-commits] CVS: sbcl/src/code early-fasl.lisp,1.34,1.35 list.lisp,1.17,1.18 seq.lisp,1.46,1.47 target-sxhash.lisp,1.9,1.10

[Christophe R. <cs...@ca...>]

Alexey Dejneka <ade...@co...> writes:

> Christophe Rhodes <cs...@ca...> writes:
>
>> Alexey, would it interfere with your CAST branch
>
> Don't bother, it's a long story. Just DTRT.
>
>> if we
>> just papered over the ENDP problems by turning it into a full
>> DEFTRANSFORM, rather than a source transform?
>
> Does it help? Or are you going to enable it only for (< SAFETY 3)? I
> also think that source transforms are faster.

I think it'll help, because it'll be automatically protected by ENDP's
DEFKNOWN.  Hm, no, it would be automatically protected by ENDP's
defknown if it were changed to read
  (defknown endp (list) boolean (foldable unsafely-flushable movable))

No argument about the "faster" bit, though...  and it is only papering
over the cracks, as our experiences with + and MAX show.

Cheers,

Christophe
-- 
http://www-jcsu.jesus.cam.ac.uk/~csr21/       +44 1223 510 299/+44 7729 383 757
(set-pprint-dispatch 'number (lambda (s o) (declare (special b)) (format s b)))
(defvar b "~&Just another Lisp hacker~%")    (pprint #36rJesusCollegeCambridge)