On 30 May 2013, at 18:21, Antonia Horincar <[email protected]> wrote:

> On Thu, May 30, 2013 at 4:25 AM, Olemis Lang <[email protected]> wrote:
>> On 5/29/13, Antonia Horincar <[email protected]> wrote:
>>> Hi,
>> 
>> :)
>> 
>>> On Tue, May 28, 2013 at 9:54 PM, Olemis Lang <[email protected]> wrote:
>> [...]
>>>> 
>>>> Yes , during the last few days I've been thinking of ways to improve
>>>> RestOnTrac plugin and I'm working now (and will be suring the next few
>>>> days) towards version 2 of the API . This means you'll get these
>>>> answers asap , and I'll be accepting both feedback and patches based
>>>> on your work .
>>>> 
>>>> What will be your starting point ? ... so that I can manage priorities
>>>> ...
>>> 
>>> Thank you for your reply. Since I still don't know very much about
>>> RestOnTrac's structure, I can't tell for sure what will be my starting
>>> point for the back-end part.
>> 
>> I meant to ask this : Will you start with embedding ticket data in
>> external sites ? or maybe some other data hosted by Bloodhound ? ...
>> which one ?
> 
> Well, I can start with any of them. Do some of the objects have
> priority over some other ones? (e.g. tickets over versions?)

I think this is a reasonable order of priority:

Ticket > Milestone > Version > Product > Component > Source > Everything else

Which part you prefer to tackle first is up to you.

> 
>> reports
>>> This is why I would appreciate some
>>> explanation on the plugin's internal structure.
>> 
>> I'll provide these asap . Nevertheless in a few words , API v2 will be
>> very similar to django-piston i.e. a Resource class with methods for
>> CRUD operations , and routes definitions in the fornt-end
> 
> Great, thanks a lot!
> 
>> 
>> 
>>>> PS: I've also been thinking of many other nice improvements too ... in
>>>> the fridge ...
>>>> ;)
>> [...]
>>> 
>>> Also, I want to share what Shiva Teja (also a member of this mailing
>>> list) suggested to me yesterday. He wanted to share with me his idea
>>> on implementing the back-end of the project, which implied using
>>> IRequestHandler
>> 
>> The RestModule class already implements IRequest handler to process
>> HTTP requests ...
>> 
>>> for dealing with web requests and the TracTicketSystem
>>> class to get the content of a specific ticket.
>> 
>> ... and (will) provide wrappers and helpers to reuse (tested)
>> BloodhoundRPC handlers retrieving data hosted by the environment (i.e.
>> not just tickets), assert permissions , pluggable API authentication
>> methods , etc ...
>> 
>>> Can anyone give some
>>> feedback on this, as I want to make a comparison between this approach
>>> and the one involving RestOnTrac, and choose the best one?
>> 
>> You could do everything from scratch , or reuse REST and/or RPC
>> backend . I'll prioritize working and testing ticket REST endpoint of
>> the API and come back to you with further details .
> 
> I guess I will use the existing implementation of IRequestHandler for
> processing HTTP requests, it wouldn't make sense to implement it from
> scratch...
> 
>> From the perspective of API consumers you'll retrieve the (XML | JSON
>> | YAML | ...) representation of ticket 123 by issuing an HTTP GET
>> request to /api/ticket/123
>> 
>> [...]
>> 
>> --
>> Regards,
>> 
>> Olemis.
> 
> Thanks,
> Antonia

Reply via email to