Re: svn: /php/php-src/trunk/Zend/tests/traits/ error_001.phpt error_002.phpt error_003.phpt error_004.phpt error_005.phpt error_006.phpt
error_007.phpt error_008.phpt error_009.phpt

Hi Felipe: Hope you don't mind that I post it on the list, think the discussion should be properly archived. On 06 May 2010, at 18:30, Felipe Pena wrote: > Hello, > > 2010/5/6 Stefan Marr <php@stefan-marr.de> > Hi Felipe: > > On 05 May 2010, at 22:00, Felipe Pena wrote: > > One question... Looking at the grammar, I got a doubt... Is actually intended to allow > > constructions like 'trait foo extends ...' and 'trait foo implements ...' ? (I > > can't see an example like that on the RFC) > > The idea is that traits form its own hierarchy of composition, and since they can be composed > from traits itself, it should not be necessary, to allow inheritance. > > So, no, extends and implements should not be allowed. > The idea of allowing interfaces was rejected in the first big discussion. > Interfaces belong to classes, and traits are solely meant to provide implementation. > Since you can exclude methods, it does not fit well with the notion of interfaces. > > There are a number of things which are not quite clear in the grammar at the moment. > My idea was that I just reuse the class grammar, and add exceptions for the situations which > are not allowed. > Mostly to avoid duplicating the relevant grammar parts. > > Do you think it would be better to do it the other way around, and use a dedicated grammar for > traits? > > We could use a dedicated grammar or make the properly checks in the compiler. Both solutions > looks good for me. > > I got a new question for you... Shouldn't we don't allow the implementation of magic > methods and constructors in Traits? Hm, well, don't know, but I do not really see a reason to disallow magic methods. There are definitely use-case for something like __toString. Constructors are a complicated topic on its own. But in general, the initialization part might also be reusable. So, what would be your argument why they might be 'harmful'? Best regards Stefan > > -- > Regards, > Felipe Pena -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525