Re: [RFC] Reflection annotations reader

2013/1/6 Yahav Gindi Bar <g.b.yahav@gmail.com> > On Mon, Jan 7, 2013 at 12:41 AM, Marco Pivetta <ocramius@gmail.com> wrote: > > > > > I think that our work is to isolate each annotation so it'll be easy to > >> access, then, it'll be easy enough to write the code that creates > "complex > >> annotations", such as constructors and so on, in userland. > >> > > > > In fact, there's probably no need (now) to go on and build a full > > annotation reader that instantiates classes and does fancy stuff as we > > currently do in doctrine/common. > > A fast parser is more than enough I suppose. That's our bottleneck > > (besides scanning files). > > > > > > Marco Pivetta > > > > http://twitter.com/Ocramius > > > > http://ocramius.github.com/ > > > > So the problem is the syntax which is difference? > When wrote this RFC, I just though about basic cases... > > Though I agree with you that the main problem is the syntax. > We can extract the entire doc-comment and only isolate between annotations, > so doc-comment like: > /** > * @Route("/") > * @ORM(Key="foo") > * @var string > */ > > Will be : array( 'Route("/")' => "", > 'ORM(Key="foo")' => "", "var" => > "string" ) > But the question is it really worth it, since you'll probably need to > create some sort of "sub-parser" that uses this isolated annotations array > and apply on them your syntax. > As a suggestion, that should cover most (all?) cases: The identifier could be defined as "between @ and the first non-alphanumeric character" (it should probably allow some special like "/", or "\" to allow namespace-like annotations). @Route("/") would be array( 'Route' => '("/")'. Now a secondary parser only needs to take care about ("/"), but for example it can already directly test, whether or not the annotation exists. > > That's being said, if we'll see performance improvements, I really think > that it's a good solution to start with, since because its not an > Attributes, I don't think that we should dictate the syntax for each > application. Each application will get the doc-comment annotation and will > be able to apply on it its own syntax and fancy stuff... I think that it's > the best solution because of BC too. > To throw that in: Multiline-annotations must be taken into account too :) Regards, Sebastian > > What do you think? > -- github.com/KingCrunch