Re: [apps-discuss] more feature requests, was: JSON patch: "test" operation

[Julian Reschke <[e-mail]>]

On 2011-12-30 06:24, Paul C. Bryan wrote:
> On Thu, 2011-12-29 at 16:40 +0100, Julian Reschke wrote:
>> Hi there,
>>
>> in discussions in Apache Jackrabbit space, two more features have been
>> mentioned as potentially useful:
>>
>> 1) The ability to send additional data along with the actual patch; such
>> as a plain text string describing the change (think "commit" message),
>> or user information.
>
> I've had a few thoughts here.
>
> First, I think commit messages should probably be out of scope for JSON
> Patch specifically. That said, we should allow the ability to build on
> the specified format, allowing for extensions, including additional
> (more domain-specific) operations.
>
> I'm generally against the idea of using out-of-band metadata such as
> header fields in an HTTP request, mostly because it ties metadata to the
> transfer protocol rather (what I believe rightly should be tied to the
> document).
>
> I'm interested in feedback on the possibility that I add text stating
> that if an operation cannot be determined for a given patch object, then
> an implementation should ignore it. Thoughts?
> ...

That's exactly the question we need to answer; is the extensibility 
model "must understand" or is it "must ignore"? In the latter case, 
people can define their own operators, such as "commit-message". But of 
course it would make it harder to define new operators later on.

If JSON had a comment syntax there would be a simple, although a bit 
hacky way to extend the format. But it does not...

Best regards, Julian