GNU bug report logs - #69864
30.0.50; Problem with equal hash tables

Previous Next

Full log

View this message in rfc822 format

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Gerd Möllmann <gerd.moellmann <at> gmail.com>
Cc: 69864 <at> debbugs.gnu.org
Subject: bug#69864: 30.0.50; Problem with equal hash tables
Date: Sun, 17 Mar 2024 16:38:19 -0400

Gerd Möllmann [2024-03-17 19:56:24] wrote:
> Gerd Möllmann <gerd.moellmann <at> gmail.com> writes:
>> (unsigned char *) data = 0x00007fb1f8a41978 "cl--generic-build-combined-method"
> Or, and maybe more plausible, someone modifies the hash-table keys used
> in the function above...

Lessee...

    (defun cl--generic-build-combined-method (generic methods)
      (let ((f
             (with-memoization
                 ;; FIXME: Since the fields of `generic' are modified, this
                 ;; hash-table won't work right, because the hashes will change!
                 ;; It's not terribly serious, but reduces the effectiveness of
                 ;; the table.
                 (gethash (cons generic methods)
                          cl--generic-combined-method-memoization)
    [...]

I suspect The Right Way would involve flushing the elements of the cache
when the generic is modified, or otherwise changing the hash-table index
so it doesn't include `generic` at all, or only the part that does
matter (as the code points out later on, currently this arg is not used
for much).

FWIW, I've never really measured the effectiveness of this cache.
It'd be worthwhile to do it at some point, because there's a good chance
its hit rater is very low (e.g. because its `:weakness` makes it lose
its info before we need it).


        Stefan

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.