perl bless/overload performance problem on RHEL
nick at ccl4.org
Wed Sep 3 10:51:48 BST 2008
On Wed, Sep 03, 2008 at 10:43:48AM +0100, Matt S Trout wrote:
> On Mon, Sep 01, 2008 at 10:16:42AM +0100, Nicholas Clark wrote:
> > What RedHat did, and why - your guess is as good as mine. But anyone
> > commenting about the source, cause, or intent of code changes to the
> > Perl core itself that is in disagreement with what I typed in
> > http://use.perl.org/~nicholas/journal/37274 is wrong.
> They originally merged your patch to negate a Class::DBI crash, since
> ::HasBeenDeleted doesn't have overloading but Class::DBI itself stringifies
> out of the box.
> The irony of the fact that this then went on to fuck over performance in
> DBIx::Class on those perls was not lost on me.
Although (as you know, but not everyone reading this might) my intent was
never to hurt performance anywhere, and it peeved me mightily that they
never considered that this inadvertent performance hit might actually be
a bug that I wasn't aware of, rather than a deliberate choice.
Hence why I fixed it. But the test cases I was aware of weren't the whole
story (but again, if that was reported to them, they either failed to
realise, or failed to pass it on), hence why I fixed the other test cases.
And learning from my mistakes, actively went out to find idioms in everyone
else's code that would fail.
(gah, is looking at average CPAN code painful)
They still haven't made any communication, even now.
More information about the london.pm