|
From: | Paul F. Dietz |
Subject: | Re: [Gcl-devel] New random tester failures |
Date: | Sat, 11 Dec 2004 07:05:49 -0600 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.2) Gecko/20040803 |
Camm Maguire wrote:
Indeed. OK, took a stab at a more general solution in CVS now. There is a quite loquacious compiler note for now describing the reasoning whereby the compiler inserts type declarations into let and let* bodies, and bodies following lambda expressions, whether the bindings are introduced by macros or not. This can be expanded upon further, primarily through careful proclamation of our existing functions. >2k random tester iterations pass,
I've run 91K iterations of the random tester overnight with terms of size (1+ (random 200)). I've now changed the random tester's probabilities to make LET/LET* forms more often and am running it again with terms of size (1+ (random 1000)). No failures yet (I reduced the chance of setf/incf/etc. so the incf/decf bug has not yet been encountered.) Paul
[Prev in Thread] | Current Thread | [Next in Thread] |