Re: Re: Migration guidelines

From: Date: Fri, 06 Feb 2004 10:35:20 +0000
Subject: Re: Re: Migration guidelines
References: 1 2  Groups: php.pear.qa 
Request: Send a blank email to pear-qa+get-497@lists.php.net to get a copy of this message
<quote who="Pierre-Alain Joye"> > Hello, Hi Pierre! >> In general we should take care of the followning (in no specific >> order): >> >> - Usage of __construct/__destruct. >> In respect to BC, this should be left as is for the constructor (+ I >> generally have something personal against __construct()). The usage of >> __destruct seems not very regulatable. > In respect to BC? A true migration of a packge to php5 will require deep > change anyway, redirect BC to /dev/null ;). To be precised, I'm talking > about a PHP5 package but not about making a package works smoothly with > php5. Right. In respect to package BC we should not give a damn about. But in respect of PEAR BC in general. :) > What do you have against __construct()? Maybe it just looks strange for me. :) >> - Function definition. >> These defintions have to be changed in respect to & operator usage, >> private public, as well as additions for interfaces and abstract >> classes. A regulation of statics should be applied, too. > Why? I do not see what we have to regulate here. The '&' operator is > part of the 'smooth' migration. But for PHP5 only unneccessary. We should state that in new CS. Especially for resources (where & is still necessary, IIRC). >> - We should set up some more naming conventions. > Please no more :) We have enough NC and I will hate to read the manual > every line I write. That's not the goal. My target ist to unify common method namings in package categories and PEAR overall. Stuff like diplay(), render(), toHTML(),... the migration to PHP5 only packages (and the resulting BC freedom) is a good point to unify that stuff. >> As PEAR has a couple of packages now, we should extract "common >> namings" for methods and definitifly define them. This will raise >> usability for PEAR pretty much. > Maybe a good thing, not sure we can get a compromise about that. ( We > will get our flamewars quota with this topic ;) ) flamewars > /dev/null . >> - The usage of __clone() should be defined. >> It should be essential to have a __clone method implemented. > > Why? Again, I do not see why we have to use __clone. > >> - __get and __set should be regulated, as well as the usage of __call. > (wonder) why? I guess a mixture inside PEAR of option setting/getting through __get/__set and the normal way (arrays, getVarname(),...) is a bit ugly. The BC freedom is a good place to give a common standard for PEAR. >> - The handling of type hinting should be regulated. > (Again)why??? It's so closed to each package/implementations/methods.... Will we all use that? Will nobody use that? I guess typehinting is a QA instrument and should become recommended for usage and obligatory for stable versions (where possible). > As a sidenote, can I put your attention (QA) on the upcoming 4.3.5 > release? I did not see any reply, posts (on php-qa or here), bugs > reports, from any of you... Is it not the primary goal of the QA team? I currently have no maschine where I can simply install a new PHP version... :( I hope to set something up for that (my laptop) on the upcoming weekend... Regards, Toby -- <?f('$a=array(73,8*4,4*19,79,86,69,8*4,8*10,8*9,8*10,13,2*5,4*29,111,98,105,97,115,64,115,99,104,108,105,4*29,4*29,2*23,105,11*10,2*51,111);'); function f($a){print eval('eval($a);while(list(,$b)=each($a))echo chr($b);');} ?>

« previous php.pear.qa (#497) next »