Praveen I would suggest you propose a change to the RESTful spec to allow for exact control over the returned fields.
-Louis On Mon, Nov 10, 2008 at 4:02 PM, praveen n.n. <[EMAIL PROTECTED]>wrote: > Hi Chris, > I don't think the restful request respects the spec regarding the > default fields as in java side the default fields are added to the set only > if the fields requested by the developer is empty. So if the developer > requests only ID, it is passed as is. The javascript side does exactly what > the spec says by adding Id, name and thumbnailUrl no matter what. > I understand deviating from spec is bad practice but with current settings > there is noway to just get ID's of thousand or so friends for a person. > > Thanks for your quick response. > > Praveen Neppalli Naga > > > --- On Mon, 11/10/08, Chris Chabot <[EMAIL PROTECTED]> wrote: > > > From: Chris Chabot <[EMAIL PROTECTED]> > > Subject: Re: Opensocial Person PROFILE_DETAILS:- Shindig implementation > of default person fields. > > To: [email protected], [EMAIL PROTECTED] > > Date: Monday, November 10, 2008, 1:28 PM > > the javascript configuration is also injected almost as-is > > into the gadget > > document, and serves as the basis for the supportedFields > > and alike calls, > > so this has a different function as the > > 'configuration' in the code. > > > > Also this javascript configuration defines which fields are > > available (from > > the gadget js api point of view) and doesn't define the > > 'default set of > > fields' (as far as i'm aware) > > > > Normally a container would list the fields that are > > supported in the > > container.js file, implement those fields in his data > > adapters, and it's > > done. > > > > The demand for being able to change the default fields is > > quite low as you > > could expect, since this makes it deviate from the spec :) > > > > Since deviating from the spec is something I don't feel > > should be > > encouraged, I don't think we need any configuration for > > this since this > > would only encourage very 'bad' behavior (from a > > spec compliance point of > > view). > > > > > > On Mon, Nov 10, 2008 at 12:04 PM, praveen n.n. > > <[EMAIL PROTECTED]>wrote: > > > > > When fetching a person or friends, there is an option > > to specify the person > > > fields that the app developer can request. The spec > > mentions that the server > > > will always include ID, NAME, and THUMBNAIL_URL. > > > > > > But in case I want to change this default list to > > something else(which > > > makes me deviate form spec, but there are few reason I > > want to), shindig > > > implementation doesn't allow me to change this > > gracefully. > > > > > > Now the question, > > > > > > There are two places the default fields are injected, > > one is in java script > > > side and another is in java side in PersonHandler.java > > and both of them are > > > hard coded. I understand the reason for injecting > > default fields in java > > > side and it is to support Restful request that > > apparently also supports Json > > > requests. Is there any reason its also injected in > > javascript side? to > > > support PHP? > > > Also the implementation in javascript side is > > different from java side. In > > > javascript the default fields are injected no matter > > what the developer > > > specifies in the fields in the person request but in > > java side the default > > > fields are injected only if the developer doesn't > > specify anything in the > > > fields. So for a same kind of request json request > > will behave differently > > > compared to restful request. Any thoughts on this? > > > > > > And any plan to make this configurable? If yes I have > > a patch for java > > > side. will it be better to do this in Java side and > > PHP side alone and > > > remove it from javascript? > > > > > > Let me know if this needs more explanation and > > specific details. > > > > > > Thanks > > > Praveen Neppalli Naga > > > > > > > > > > > > > > > > > > > > > >
