Re: dwbp-ISSUE-213 (BP_REST): Changes on BP about REST APIs [Best practices document(s)]

Hi Annette,

My view on this matter:

The audience for this working group has no message of REST and HATEOAS 
being best practises. REST and HATEOAS could be seen as a set of best 
practises on their own. Referencing this from an API BP would seem a bit 
meta to me.

Instead I would suggest taking it a step further. When we combine REST, 
HATEOAS, Linked Data and HTTP for an Open Data API thoroughly, you'd 
come up with Hydra [1] (a community group [2] at W3C). Wouldn't it be 
better to create a best practise that says you need to weave your data 
documents together with Hydra?

[1] http://www.hydra-cg.com/
[2] https://www.w3.org/community/hydra/

Kind regards,

Pieter

On 16-11-15 20:35, Annette Greiner wrote:
> Hi Yaso,
> Yes, I think we have covered the needs of those who are not publishing 
> with an API pretty well (possibly too well, but that's another 
> discussion).
> I wouldn't want to include HATEOAS in the title of the BP for using an 
> API. The BP should be about what the HATEOAS and pragmatic camps have 
> in common. I think it's important to mention both the pragmatic REST 
> approach and the HATEOAS approach as distinct implementation options.
> The current BP 21 has a lot of issues that we've been discussing 
> improvements on for a while. I don't think what we have in the google 
> doc is perfect, but I do think it at least addresses the issues 
> raised. How do you feel about swapping that in and then continuing to 
> improve it?
> -Annette
>
> On 11/16/15 4:19 AM, Yasodara wrote:
>> Hi all
>>
>> I exchanged a few ideas on this BP with Newton before he raised the
>> issue. Although I had written this BP, I agree that recommending RESTful
>> architecture may cause some confusion. The discussion on what is a
>> RESTful API, or what can be the ideal architecture, is beyond our scope,
>> imho. If we enter this issue we may have to discuss, for example, the
>> "maturity level" [1] proposed for rest APIs..
>>
>> Maybe, for the sake of the document, is better reconsidering
>> recommending the REST (with this exactly word) in the top of the BP.
>>
>> But I would mention HATEOAS **instead**.
>>
>> I agree with Newton that "it would be interesting suggest the use of an
>> REST API as an approach to implementation of the BP" instead of a
>> recommendation."
>>
>> @annete, about mentioning other use cases that doesn't use APIs.. . I
>> think we already covered that. Don't we?
>>
>>
>> cheers,
>> yaso
>>
>>
>> http://martinfowler.com/articles/richardsonMaturityModel.html
>>
>>
>>
>>
>>
>>
>> On 11/16/2015 09:59 AM, Bernadette Farias Lóscio wrote:
>>> Hi Annette,
>>>
>>> I agree with you! I think using API is just one of the possible ways of
>>> publishing data on the Web.
>>>
>>> cheers,
>>> Bernadette
>>>
>>> 2015-11-13 20:02 GMT-03:00 Annette Greiner <amgreiner@lbl.gov>:
>>>
>>>> My reason for asking is that I think it's important to consider 
>>>> possible
>>>> use cases for doing something other than publishing an API. I 
>>>> realize it
>>>> doesn't meet your criteria for being "webby", but there are people 
>>>> who use
>>>> the web in other ways, and they may have good reason for doing so. 
>>>> I don't
>>>> want to be dismissive of them.
>>>> -Annette
>>>>
>>>> On 11/13/15 12:57 PM, Erik Wilde wrote:
>>>>
>>>>> On 2015-11-13 19:38, Annette Greiner wrote:
>>>>>
>>>>>> I would not recommend the use of SOAP to anyone these days. I do
>>>>>> recommend the use of REST principles, even in backends that are 
>>>>>> not set
>>>>>> up as an API per se, because they make it easy to add an API 
>>>>>> later. Is
>>>>>> it always a best practice to make an API available? I'm not sure.
>>>>>>
>>>>> well, if it's "data on the web" i'd assume at the very least there is
>>>>> some remote access in the overall picture, right? if that;s the 
>>>>> case, then
>>>>> it can either be the FTP/SOAP-style "let's serve some files via 
>>>>> whatever
>>>>> transport channel we have", in which case i'm not sure why the 
>>>>> title even
>>>>> mentions the web, or it could be a way of doing this webby, in 
>>>>> which case
>>>>> your data model is your API, essentially.
>>>>>
>>>>> cheers,
>>>>>
>>>>> dret.
>>>>>
>>>>>
>>>> -- 
>>>> Annette Greiner
>>>> NERSC Data and Analytics Services
>>>> Lawrence Berkeley National Laboratory
>>>>
>>>>
>>>>
>>>
>

-- 
+32486747122
Linked Open Transport Data researcher
UGent - MMLab - iMinds

Board of Directors Open Knowledge Belgium
http://openknowledge.be

Open Transport working group coordinator at Open Knowledge International
http://transport.okfn.org

Received on Monday, 16 November 2015 19:53:06 UTC