On Fri, Oct 17, 2014 at 06:40:42PM +0200, Steffen Mueller wrote:
> On 10/17/2014 06:33 PM, Father Chrysostomos wrote:> >I'm trying to extend an optimisation (see the message> >entitled 'list+pushmark optimisation for review' or> ><2014...@lists-nntp.develooper.com>), and to do> >this I need to renumber most of the sequence numbers tested for in> >f_map.t. f_map.t is full of version-specific checks, so keeping it> >working with older perl versions would be duplicating the entire> >script. This is madness. May I just remove those version checks?> > Please do. Been there! This stuff isn't dual-lived as far as I know, and it> makes very little sense to backport B, IMNSHO.> > But I'd like to hear if anyone else has thoughts on this. NB: Maybe get in> touch with Reini off-list since his opinion on this might actually be> relevant.
Two years ago I did a bunch of work to make B stuff work under older perl
versions. I came to the conclusion that it was an unsustainable task, and
that we should not bother making it CPAN-isable, nor bother making it
suitable for maint braches - they can get their own per-branch mods rather
than trying to sync across branches.
(See e.g.
git log --author=davem 32b17be10822fe517336 ext/B
)
--
Any [programming] language that doesn't occasionally surprise the
novice will pay for it by continually surprising the expert.
-- Larry Wall