Bug in sxhash

Bug in sxhash

[Kim F. Storm]

Is sxhash really supposed to ignore the cdr of a cons?

(sxhash '(a . 1))  => 57
(sxhash '(a . 2))  => 57
(sxhash '(a . 3))  => 57

It works ok for lists:

(sxhash '(a 1))  => 913
(sxhash '(a 2))  => 914
(sxhash '(a 3))  => 915


Below is a proposed patch which gives this result:

(sxhash '(a . 1))  => 913
(sxhash '(a . 2))  => 914
(sxhash '(a . 3))  => 915

[i.e. there is no difference between lists and cons,
but at least it isn't completely stupid]


*** fns.c	06 Feb 2006 18:21:46 +0100	1.407
--- fns.c	21 Mar 2006 16:57:07 +0100	
***************
*** 5114,5119 ****
--- 5114,5125 ----
  	hash = SXHASH_COMBINE (hash, hash2);
        }
  
+   if (!NILP (list))
+     {
+       unsigned hash2 = sxhash (list, depth + 1);
+       hash = SXHASH_COMBINE (hash2, hash);
+     }
+ 
    return hash;
  }
  

--
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Bug in sxhash

[Richard Stallman]

I think your change is desirable--please install it.

Re: Bug in sxhash

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

> I think your change is desirable--please install it.
Done.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk