On Thu, Aug 25, 2011 at 10:39 AM, Zeev Suraski <[email protected]> wrote:
>> -----Original Message-----
>> From: Pierre Joye [mailto:[email protected]]
>> Sent: Thursday, August 25, 2011 11:31 AM
>> To: [email protected]
>> Cc: Stas Malyshev; [email protected]
>> Subject: Re: [PHP-DEV] PHP 5.3.8 Released!
>>
>> On Thu, Aug 25, 2011 at 3:51 AM, [email protected]
>> <[email protected]> wrote:
>>
>> > BTW. we could really do with a searchable archive of php.dev + internals...
>> > - It's pretty difficult to find out if this was ever discussed before...
>>
>> Again, it was discussed already and the argument of using instanceof was used
>> to deprecate is_a (along other arguments). I see no point to argue again about
>> that. It was a mistake to change it again in 5.3, Zeev realized the use in this
>> exact case, let move on now. Revert that in 5.3 and do it only in 5.4.
>
> Just so that my position is clear on the matter:
> - I still think that is_a() is working properly the way it does now, after the change.
> - I think the code in isError() is wrong, and should be fixed.
> - Had we (the collective we) known that fixing is_a() would result in such breakage, it would
> have probably been wise not to fix it in 5.3.x, and wait for 5.4 for this purpose.
> - Given that we've already done it - I wouldn't revert it. Fix it in PEAR..
> That's the only way to create something that works across all versions of 5.3.x.
I won't battle endlessly to get that fixed back again (you will always
find smtg else to keep that thing in anyway ;-) but it does confirm
what I said earlier about changing behaviors in patch releases. This
is something we must deal with much more carefully. And using x.y.0
tests as base is a good start, also the BC breakages are now covered
by the RFC, let stick to that for 5.4.x.
Cheers,
--
Pierre
@pierrejoye | http://blog.thepimp.net | http://www.libgd.org