develooper Front page | perl.perl5.porters | Postings from March 2015

Re: OP_SIGNATURE

Thread Previous | Thread Next
From:
Karl Williamson
Date:
March 5, 2015 16:55
Subject:
Re: OP_SIGNATURE
Message ID:
54F88A59.6000006@khwilliamson.com
On 03/05/2015 09:15 AM, H.Merijn Brand wrote:
> On Thu, 5 Mar 2015 17:08:18 +0100, demerphq <demerphq@gmail.com> wrote:
>
>> On 5 March 2015 at 16:23, Zefram <zefram@fysh.org> wrote:
>>> Dave Mitchell wrote:
>>>>                                              Since other people are making
>>>> assertions that it should be possible to get nearly as good performance as
>>>> OP_SIGNATURE with a combination of more generic ops,
>>>
>>> Who's asserting that?  I haven't, and I wouldn't because, like you, I
>>> wouldn't be convinced of it without a fairly specific design.  Indeed,
>>> I have already explicitly acknowledged that it would be difficult for
>>> anything else to equal the signature op for speed.
>>>
>>>>                                                      the onus is *you* to
>>>> come up with a specific proposal, which I can then either accept, tweak, or
>>>> demonstrate to be slower.
>>>
>>> My objections to the signature op are not based on any assertion that
>>> its performance is suboptimal, so demonstrating that another design is
>>> slower (or even that all possible other designs are slower) would not
>>> refute my objection.  You seem to be projecting your monomoniacal focus
>>> on speed onto those who argue againt the signature op.  It's coming
>>> across as a straw man.
>>
>> To me it is coming across as frustration that you aren't acknowledging
>> the importance of speed to the vast majority of users.
>>
>> And I share that frustration.
>>
>> I don't believe signature pluggability is a very important goal in
>> comparison to performance improvements to subroutine calls, and I
>> don't accept the premise that it is better to do nothing than it is to
>> add OP_SIGNATURE. I think that you can have what you want once you
>> build it, and that adding OP_SIGNATURE now shouldn't get in the way of
>> it.
>
> Hear, hear!
>
> +1
>

+1

Given that this is an experimental feature, I don't see any validity in 
saying it's better to do nothing.  Regardless of what the eventual API 
comes out as, it's just experimental, now.

Thread Previous | Thread Next


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