Re: ISSUE-58: the simple solution to inlined membership - ISSUE-45

On 18 May 2013, at 18:28, "Wilde, Erik" <> wrote:

> hello richard.
> On 2013-05-18 10:06 , "Richard Cyganiak" <> wrote:
>> No, the reason was that POST is HTTP's catch-all extension point for
>> "other" operations, and it would be bad to "use up" that extension point
>> on all LDP-managed resources. LDP implementers should have the liberty of
>> using POST to an LDPR for other things beside append.
> my favorite topic! i have used up all HTTP methods, now what? ;-) sorry,
> but i really cannot resist...
> you cannot "use up" all HTTP methods. you model all the interactions with
> your resources as self-describing interactions. for each of those
> interactions, you look at whether they are safe and idempotent and pick
> GET if they are, they delete things and pick DELETE if they do, they are
> idempotent but not safe and pick PUT, and otherwise you pick POST.

Except you can't use POST because your web framework supports LDP, and in LDP a POST to any resource means "append triples to this resource" (if we go with Henry's plan).

I don't want to restrict the use of POST to only this one purpose in LDPRs.


> the
> fact that early writings about HTTP/REST hardcoded the operation
> names/semantics is more historical at this point, and REST has moved on to
> focus on the *interaction semantics* when picking a method, because that's
> what matters when it comes to handling the HTTP methods in the web's
> fabric.
> long story short: don't worry, you can never "use up" all HTTP methods.
> when you use POST to append, that's fine and just make sure it's
> self-describing. when you add other interactions that also use POST,
> there's no conflict because these interactions should be self-describing
> as well.
> in a nutshell, it's the same as having various "submit" buttons on an HTML
> form: they all go to the same form/@action URI, but since you're pressing
> a different button, your interacting with the resource in a different way.
> cheers,
> dret.

Received on Saturday, 18 May 2013 17:36:42 UTC