W3C home > Mailing lists > Public > public-expath@w3.org > February 2015

Re: Common HTTP Model

From: Christian Grün <christian.gruen@gmail.com>
Date: Sat, 28 Feb 2015 13:04:07 +0100
Message-ID: <CAP94bnOhHfGsfz4--4EQkOYhqgmS640zU18S6+ksbx-2sjSXiA@mail.gmail.com>
To: Adam Retter <adam@exist-db.org>
Cc: Florent Georges <fgeorges@fgeorges.org>, EXPath <public-expath@w3.org>
Hi,

As the HTTP Model is simply a description of how HTTP requests and
responses can be represented via XML, there may be no need to make it
optional.

I agree with Adam that there may not always be a requirement for
functions accessing this model, so it would be just consequent to
treat them as optional extension. In RESTXQ, for example, the HTTP
Model is used to build server responses, and those results will
usually never be requested from within a function. Instead, it is the
RESTXQ implementation that parses this XQuery response result and
creates the real HTTP response.

Personally, I can life with both solutions: Define two specs or keep
them together. In terms of naming things, it might be more consequent
to split them. Right now, on top of [1] it says "HTTP Model. EXPath
Candidate Module 11 January 2014". But I believe that a model is
nothing like a "module". A better naming could be:

* HTTP Model. EXPath Candidate ..
* HTTP Module. EXPath Candidate Module

In a broader scope, we could even think about rephrasing the subtitles
of each "EXPath Module" to "EXPath Recommendation". Aftre all, in my
points of view, features like the packaging system are much more than
plain modules.

What do you think?
Christian

[1] http://expath.org/spec/http/editor




On Fri, Feb 27, 2015 at 11:30 PM, Adam Retter <adam@exist-db.org> wrote:
> Florent,
>
> I still think I have a problem with this spec defining two things,
> i.e. a model and functions. I would rather see two specs, in this way
> I or anyone else can choose to implement either or both of the model
> and functions. It makes no sense to me to conflate the concept of
> modelling with a function library. If you must you could split it into
> two parts where both are optional, however I think two separate spec
> documents is preferable even if they had to share the same namespace.
>
> Also... just wondering why the URL is suffixed '/editor', is there
> some sort of significance to that?
>
> On 26 February 2015 at 19:34, Florent Georges <fgeorges@fgeorges.org> wrote:
>>   Just a quick email to point out there is already an editor draft at
>> http://expath.org/spec/http/editor.  I still have pending changes I
>> need to commit, I'll have a look as soon as I can.
>>
>> --
>> Florent Georges
>> http://fgeorges.org/
>> http://h2oconsulting.be/
>>
>>
>> On 25 February 2015 at 13:30, Christian Grün wrote:
>>>> Before the HTTP Client Module hits 1.0, it would be very nice if we
>>>> could define a common XML HTTP Model for HTTP Requests/Responses
>>>> (excluding bodies initially) which could live outside of the
>>>> http://expath.org/ns/http-client namespace.
>>>
>>> Yes, I believe that an own spec for defining the HTTP model would make
>>> sense. We have the HTTP Client Module and RESTXQ today, and we may
>>> have other extensions in future that could be based on this model.
>>> Maybe it even simplifies the finalization of the HTTP Client Module,
>>> as it its contents will be more focused on the actual task.
>>>
>>> +1 from me,
>>> Christian
>>>
>
>
>
> --
> Adam Retter
>
> eXist Developer
> { United Kingdom }
> adam@exist-db.org
> irc://irc.freenode.net/existdb
Received on Saturday, 28 February 2015 12:04:55 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:47:39 UTC