Venerable monks,
someone just told me that their perl (v5.8.1-RC3/darwin) stored hashes differently between different invocations of a program. Now, while we certainly can't rely on the 'order' in which hash entries are stored, I was assuming (until now) that the 'order' was at least consistent.
But it turns out that multiple calls to
$ perl -e '%a=(1=>"a",2=>"b",3=>"c"); print %a,"\n"'
produce different output on this platform, like this:
2b3c1a
3c1a2b
...
Does anybody know if there's a random generator in the hash algorithm of some perl implementations?