Re: Re: PEAR::Date broken (Was: [PHP-CVS] cvs: php-src(PHP_5_2) /ext/date php_date.c php_date.h)

On 7/18/06, John Coggeshall <john@coggeshall.org> wrote:

On Tue, 2006-07-18 at 20:25 +0200, Edin Kadribasic wrote: But breaking even a few or "both" is still reckless, irresposible behaviour when all that is needed to fix the breakage is rename the class. Espacially because of all the bad publicity we get for breaking backwards compatibility for no good reason. In my experience, we've received an equal if not greater amount of negative publicity from a lack of consistent naming conventions. PEAR is a library of functionality developed on top of PHP, and therefore PHP has the first say to what to name what and everyone else moves around that. I also think that as PHP moves from a strictly procedural language to an ever growing OO arm, IMHO it isn't unreasonable to lay claim to certain core class names. Moreover, I think this whole flame war is counterproductive for us all -- can we just do a +/- vote on this and move on to more useful concerns? +1 for Date as the name.

I'm indeed in favour of Date. But what I'm not in favour is: - to add/enable features not approved right before releases (for the 3rd time, and in a row) - to start using the common names in general without a loud, "official" and preemptive warning to our users (meaning not from one minor to another) And what I cannot accept is people arguing about PEAR dictating PHP developments. We do not, but Derick did. So please spare me the endless "Oh Look! PEAR is crap and evil", it is actually unopportune. And last but not least, I fully agree with Andi and you, the needs of classes internally is growing, faster than ever before. We have to define rules or conventions. However I don't think that discussing it right before a RC release is the right time. We better have to remove the controversial part (drop the class, keep the functions) and take the time to discuss this problem, generally and not only for Date. Cheers, --Pierre