JSONP for /gadgets/metadata?

[John Hjelmstad <fa...@google.com>]

I've recently been talking with a few parties who are interested in using
/gadgets/metadata functionality via JSONP. It's understood that use of the
API this way would be somewhat limited due to URL length restrictions, but
that seems reasonable since most use cases involve retrieval of metadata for
a single gadget.
This seemed reasonable to me since gadgets themselves contain no sensitive
information. The biggest con I had in mind is that callers would need to be
careful about what UserPrefs (et al) they send as a GET.

I'm happy to do the implementation myself, but wanted to air this idea with
the list in case I'm missing something key.

Proposed impl:
/gadgets/metadata?req=<url-encoded-equivalent-of-POST>&callback=<arg.matches([a-zA-Z\.]+)>

Thoughts?
--John

Re: JSONP for /gadgets/metadata?

[Chris Chabot <ch...@google.com>]

I'm def. +1 on the concept, the meta data service is used by & very useful
for quite a few parties, making it more documented will only increase this.

On Wed, Sep 17, 2008 at 6:11 PM, Kevin Brown <et...@google.com> wrote:

> On Wed, Sep 17, 2008 at 5:27 PM, Arne Roomann-Kurrik <kurrik@google.com
> >wrote:
>
> > Is the current metadata implementation documented somewhere?  I'm working
> > on
> > a Firefox extension that could benefit from using such a service.
>
>
> No, just in source -- if we migrated it to a format closely matching the
> social data calls (JSON-RPC only at this point), most of the documentation
> would be done for us already.
>
>
> >
> >
> > ~Arne
> >
> >
> >
> > On Wed, Sep 17, 2008 at 5:06 AM, Kevin Brown <et...@google.com> wrote:
> >
> > > I'd actually like to see the metadata call reworked significantly to
> > > closely
> > > resemble the social data calls, and then we can provide JSONP support
> for
> > > all of them with some shared common code.
> > > Note that I'm *not* suggesting that social-api support the metadata
> call,
> > > simply that the metadata call can be used with the same syntax. This
> > makes
> > > it so that a client that has been authored to consume social data can
> now
> > > also consume gadget metadata.
> > >
> > > On Wed, Sep 17, 2008 at 1:42 AM, John Hjelmstad <fa...@google.com>
> > wrote:
> > >
> > > > I've recently been talking with a few parties who are interested in
> > using
> > > > /gadgets/metadata functionality via JSONP. It's understood that use
> of
> > > the
> > > > API this way would be somewhat limited due to URL length
> restrictions,
> > > but
> > > > that seems reasonable since most use cases involve retrieval of
> > metadata
> > > > for
> > > > a single gadget.
> > > > This seemed reasonable to me since gadgets themselves contain no
> > > sensitive
> > > > information. The biggest con I had in mind is that callers would need
> > to
> > > be
> > > > careful about what UserPrefs (et al) they send as a GET.
> > > >
> > > > I'm happy to do the implementation myself, but wanted to air this
> idea
> > > with
> > > > the list in case I'm missing something key.
> > > >
> > > > Proposed impl:
> > > >
> > > >
> > >
> >
> /gadgets/metadata?req=<url-encoded-equivalent-of-POST>&callback=<arg.matches([a-zA-Z\.]+)>
> > > >
> > > > Thoughts?
> > > > --John
> > > >
> > >
> >
> >
> >
> > --
> > OpenSocial IRC - irc://irc.freenode.net/opensocial
> >
>

Re: JSONP for /gadgets/metadata?

[Kevin Brown <et...@google.com>]

On Wed, Sep 17, 2008 at 5:27 PM, Arne Roomann-Kurrik <ku...@google.com>wrote:

> Is the current metadata implementation documented somewhere?  I'm working
> on
> a Firefox extension that could benefit from using such a service.


No, just in source -- if we migrated it to a format closely matching the
social data calls (JSON-RPC only at this point), most of the documentation
would be done for us already.


>
>
> ~Arne
>
>
>
> On Wed, Sep 17, 2008 at 5:06 AM, Kevin Brown <et...@google.com> wrote:
>
> > I'd actually like to see the metadata call reworked significantly to
> > closely
> > resemble the social data calls, and then we can provide JSONP support for
> > all of them with some shared common code.
> > Note that I'm *not* suggesting that social-api support the metadata call,
> > simply that the metadata call can be used with the same syntax. This
> makes
> > it so that a client that has been authored to consume social data can now
> > also consume gadget metadata.
> >
> > On Wed, Sep 17, 2008 at 1:42 AM, John Hjelmstad <fa...@google.com>
> wrote:
> >
> > > I've recently been talking with a few parties who are interested in
> using
> > > /gadgets/metadata functionality via JSONP. It's understood that use of
> > the
> > > API this way would be somewhat limited due to URL length restrictions,
> > but
> > > that seems reasonable since most use cases involve retrieval of
> metadata
> > > for
> > > a single gadget.
> > > This seemed reasonable to me since gadgets themselves contain no
> > sensitive
> > > information. The biggest con I had in mind is that callers would need
> to
> > be
> > > careful about what UserPrefs (et al) they send as a GET.
> > >
> > > I'm happy to do the implementation myself, but wanted to air this idea
> > with
> > > the list in case I'm missing something key.
> > >
> > > Proposed impl:
> > >
> > >
> >
> /gadgets/metadata?req=<url-encoded-equivalent-of-POST>&callback=<arg.matches([a-zA-Z\.]+)>
> > >
> > > Thoughts?
> > > --John
> > >
> >
>
>
>
> --
> OpenSocial IRC - irc://irc.freenode.net/opensocial
>

Re: JSONP for /gadgets/metadata?

[Arne Roomann-Kurrik <ku...@google.com>]

Is the current metadata implementation documented somewhere?  I'm working on
a Firefox extension that could benefit from using such a service.

~Arne



On Wed, Sep 17, 2008 at 5:06 AM, Kevin Brown <et...@google.com> wrote:

> I'd actually like to see the metadata call reworked significantly to
> closely
> resemble the social data calls, and then we can provide JSONP support for
> all of them with some shared common code.
> Note that I'm *not* suggesting that social-api support the metadata call,
> simply that the metadata call can be used with the same syntax. This makes
> it so that a client that has been authored to consume social data can now
> also consume gadget metadata.
>
> On Wed, Sep 17, 2008 at 1:42 AM, John Hjelmstad <fa...@google.com> wrote:
>
> > I've recently been talking with a few parties who are interested in using
> > /gadgets/metadata functionality via JSONP. It's understood that use of
> the
> > API this way would be somewhat limited due to URL length restrictions,
> but
> > that seems reasonable since most use cases involve retrieval of metadata
> > for
> > a single gadget.
> > This seemed reasonable to me since gadgets themselves contain no
> sensitive
> > information. The biggest con I had in mind is that callers would need to
> be
> > careful about what UserPrefs (et al) they send as a GET.
> >
> > I'm happy to do the implementation myself, but wanted to air this idea
> with
> > the list in case I'm missing something key.
> >
> > Proposed impl:
> >
> >
> /gadgets/metadata?req=<url-encoded-equivalent-of-POST>&callback=<arg.matches([a-zA-Z\.]+)>
> >
> > Thoughts?
> > --John
> >
>



-- 
OpenSocial IRC - irc://irc.freenode.net/opensocial

Re: JSONP for /gadgets/metadata?

[Kevin Brown <et...@google.com>]

I'd actually like to see the metadata call reworked significantly to closely
resemble the social data calls, and then we can provide JSONP support for
all of them with some shared common code.
Note that I'm *not* suggesting that social-api support the metadata call,
simply that the metadata call can be used with the same syntax. This makes
it so that a client that has been authored to consume social data can now
also consume gadget metadata.

On Wed, Sep 17, 2008 at 1:42 AM, John Hjelmstad <fa...@google.com> wrote:

> I've recently been talking with a few parties who are interested in using
> /gadgets/metadata functionality via JSONP. It's understood that use of the
> API this way would be somewhat limited due to URL length restrictions, but
> that seems reasonable since most use cases involve retrieval of metadata
> for
> a single gadget.
> This seemed reasonable to me since gadgets themselves contain no sensitive
> information. The biggest con I had in mind is that callers would need to be
> careful about what UserPrefs (et al) they send as a GET.
>
> I'm happy to do the implementation myself, but wanted to air this idea with
> the list in case I'm missing something key.
>
> Proposed impl:
>
> /gadgets/metadata?req=<url-encoded-equivalent-of-POST>&callback=<arg.matches([a-zA-Z\.]+)>
>
> Thoughts?
> --John
>