[geeklog-devel] Feedback on 1st draft of OpenID 2.0 support GSoC proposal

[Evgeniy Mikhaylov]

Hello, Dirk


> >* 5 weeks to implement llibrary.
>
> Would this include support for the Provider part? That's not entirely
> clear since your suggesting to implement that later in the project. You
> may also want to re-evaluate this estimation regarding what I just wrote in
> http://eight.pairlist.net/pipermail/geeklog-devel/2009-March/004139.html
Yes. This library will implement both parts: OpenID consumer and
provider. I will try to rewrite this part to make it more explicit. I
am going to implement library in the first part of summer. By the end
of the summer i will wirte plugin for Geeklog which will utilize this
lib to run OpenID server/provider.


> Since this a relatively isolated, non-vital feature and the task is hard
> enough, I don't see a need to support PHP 4 any more. Compare with the
> Webservices (AtomPub), which are also only available with PHP 5 for
> pretty much the same reasons.
Great.


> >Unfortunately right now Gl doesn't have test suites so i should take
> >into account probable changes in testing framework.
>
> The test framework project would come to mind (should that happen and be
> successful). But sticking with the idea of making this library
> explicitly _not_ depend on Geeklog, I'd suggest using PHPUnit from the start.
I need to make a decision about testing suite not only for the libary
:-) It seems to be hard to replace existing OpenID consumer parts
without writing test cases. I mean i will write tests for OpenID login
part of Gl and will replace existing code under control of tests. It
would be nice to have the same framework in Gl and this library
project, isn't it?



> >it will include a lot of "diving into specs" sessions. On the other
> >there are several open sources libraries which can be used as a reference.
> Just make sure not to take too much inspiration from them or we may have
> the license issue again that this library was supposed to solve in the
> first place. This should really be used as a last resort - try to do a
> "clean room" implementation as much as possible.
I am not going to copy/paste code from existing libraries, but i
always thought that i can use open source projects to see how
particular things were implemented and implement it in the same way in
my project (can i do it?). Especially if i am using project which was
written in another programming language.


> I'd like to see a few words convincing us that you're the right one for
> the task and that you won't let us down by dropping out of the project
> half-way through. Be creative ;-)
Thanks. I didn't want to post some personal data to the public mail
list. Be sure that i will add few paragraphs tagged "Why me".