develooper Front page | perl.perl5.porters | Postings from September 2012

Silver Bullet

From:
Reini Urban
Date:
September 18, 2012 12:13
Subject:
Silver Bullet
Message ID:
CAHiT=DGutxvezXpuLnXNLQLtkkqyuPpjRzZ9hY2LsReHMugd3g@mail.gmail.com
On Mon, Sep 17, 2012 at 6:33 AM, Nicholas Clark <nick@ccl4.org> wrote:
> On Tue, Sep 11, 2012 at 09:27:27AM -0500, Jesse Luehrs wrote:
>> On Tue, Sep 11, 2012 at 10:00:19AM +0100, Steffen Schwigon wrote:
>
>> > Speed matters, that's a fact. Not allowing discussion before Reini comes
>> > up with the silver bullet isn't fair either.
>
> There is no silver bullet.
>
> I think that it's accurate to say that everyone agrees that Perl 5 (as is)
> is too dynamic to optimise well.
>
> Reini's broad thrust seems to be to take the approach of toning down the
> dynamism in order to apply conventional (static language) optimisation
> techniques.
>
> That will get some speedup, but without (at least) prototyping it, it's not
> clear how much. Or how invasive the necessary language changes would be.
>
> But it's certainly not a silver bullet. The silver bullet would be a way
> to speed up all (or most) existing code on CPAN, unchanged.
>
>> I fail to see where anyone here is not allowing discussion. In fact, we
>> seem to have discussions about this pretty regularly(:
>
> Agree. We seem to have a lot of discussions about this and quite a few things
> pretty regularly. If there seems to be any concern about "not allowing
> discussion", it's more that nearly all of these discussions are mere
> talking shops, and go nowhere. It seems that many people have opinions, but
> very few actually have anything more substantive to contribute. This becomes
> immensely frustrating to the people who could actually do something more
> substantive, but seem to find all their time and enthusiasm consumed by
> dealing with e-mail.
>
> It's not "no discussion". It's "please, more code, less code-free discussion"
>
> Very much a reaction to "meetings, the practical alternative to work".

Thanks.

I thought the general compile-time optimization ideas are still in
memory from 2001,
but obviously not so.
So I'm preparing a short summary of my plans. I talked about it publicly
at YAPC's also.

Currently an early draft is https://gist.github.com/3740313
but is too short to be discussable yet.
-- 
Reini Urban
http://cpanel.net/   http://www.perl-compiler.org/



nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About