Re: Hash seed breaks 5.8.1 binary API; fix suggested
According to Alan Burlison:
> Chip Salzenberg wrote:
> >All I want is binary compatibility for 5.8.x
> One question: will your proposed change undo the security fix, the
> reason why the change was made in the first place?
No, I would break bincompat if security required it.
Fortunately it doesn't. (Apparently.)
> Also, I didn't really understand whey we needed yet another global
We need a global internal to Perl to record the hash seed for the
current interpreter instance. We don't need to export it, though,
since XS object code no longer needs to reference it.
(OTOH, we must continue to export PL_hash_seed [which will always be
zero, forever] because XS code compiled against 5.8.1 will contain
external references to it.)
Chip Salzenberg - a.k.a. - <email@example.com>
"I wanted to play hopscotch with the impenetrable mystery of existence,
but he stepped in a wormhole and had to go in early." // MST3K