Plan C Worked. And there was much rejoicing. (yeah)
According to Nicholas Clark:
> Something like this? Only with the rough edges smoothed out?
In short: Nick++
In long: Brendan O'Dea, Debian's Perl maintainer, has verified that
HTML::Parser built under 5.8.0 fails with 5.8.1, but works with 5.8.1
and the Plan C patch. Since there's just the one bug and HTML::Parser
tickled it, this is enough proof for me that Plan C works.
Brendan has uploaded the patched 5.8.1 to Debian unstable, so it will
get significant usage, but only by users who are relatively tolerant
of breakage. (Debian unstable is ideal for this kind of thing.)
However, I agree with Nick that some refinement is needed, and I'm
sure Brendan will be happy to upload again when the next iteration of
Plan C is available.
> I don't think that it really needs two set variables, given that we
> don't set the 5.8.1 hash seed anymore. Also, MJD's calculations
> convince me that the threshold for changing strategy is too conservative,
> and that it might be better to initiate a split in the store routines
> based on linked list length, rather than buckets used.
Chip Salzenberg - a.k.a. - <firstname.lastname@example.org>
"I wanted to play hopscotch with the impenetrable mystery of existence,
but he stepped in a wormhole and had to go in early." // MST3K