RE: [PHP-DEV] Unicode and fetch class
| From: | Andi Gutmans | Date: | Tue, 18 Jul 2006 00:31:45 +0000 |
| Subject: | RE: [PHP-DEV] Unicode and fetch class | ||
| References: | 1 | Groups: | php.internals |
| Request: | Send a blank email to internals+get-24452@lists.php.net to get a copy of this message | ||
OK sounds fine.
> -----Original Message-----
> From: Andrei Zmievski [mailto:andrei@gravitonic.com]
> Sent: Monday, July 17, 2006 5:28 PM
> To: Andi Gutmans
> Cc: 'Zeev Suraski'; 'Marcus Boerger'; 'PHP internals'
> Subject: Re: [PHP-DEV] Unicode and fetch class
>
> We did talk about this before, on the php-unicode list if you
> remember. The UTR#31 has a section about normalization:
>
> http://www.unicode.org/reports/tr31/
>
> I think we should see what kind of performance penalties
> identifier normalization brings with it, if any. Once we have
> the numbers, it's very easy to replace full normalization
> with simple case-folding.
>
> -Andrei
>
>
> On Jul 17, 2006, at 5:12 PM, Andi Gutmans wrote:
>
> > Is there a good reason for normalizing? I thought we are fine with
> > expecting developers to use standard form and not suffer
> performance
> > penalties. I'm not a Unicode expert so my question might be
> somewhat
> > ignorant..
> >
> >> -----Original Message-----
> >> From: Andrei Zmievski [mailto:andrei@gravitonic.com]
> >> Sent: Monday, July 17, 2006 4:57 PM
> >> To: Zeev Suraski
> >> Cc: Andi Gutmans; Marcus Boerger; PHP internals
> >> Subject: Re: [PHP-DEV] Unicode and fetch class
> >>
> >> Yes, we are. But normalization (in our case) may involve
> case-folding
> >> (for class/function names), so it's really a single
> operation. If you
> >> are asking whether we should have 2 flags for it, I don't
> believe so.
> >>
> >> -Andrei
> >>
> >>
> >> On Jul 17, 2006, at 4:48 PM, Zeev Suraski wrote:
> >>
> >>> Aren't we both normalizing and lowercasing?
> >>>
> >>> Zeev
> >>>
> >>> At 02:45 18/07/2006, Andrei Zmievski wrote:
> >>>> We are actually normalizing, but only identifiers.
> >>>>
> >>>> -Andrei
> >>>>
> >>>> On Jul 17, 2006, at 4:08 PM, Andi Gutmans wrote:
> >>>>
> >>>>> Looks OK to me. Can we call it
> >> ZEND_FETCH_CLASS_NO_LOWERCASE instead
> >>>>> of ZEND_FETCH_CLASS_NO_NORMALIZE. No biggy but I always
> think of
> >>>>> normalizing as in taking a Unicode string and normalizing
> >> it. Or is
> >>>>> that what you're actually doing here? Asking because we
> agreed we
> >>>>> won't normalize automagically in PHP.
> >>
>