It's true that in some places the response payload includes more
information than is necessary given the context. That is largely due to
the fact that the same information can be returned from multiple URLs,
as Jakub noted. So asking for all services in an Org returns the same
info as if you search for services. In the latter case the org info is
needed.
@Jakub - Regarding IDs I think that's a very good/valid point. I've
added a feature request to JIRA to include URIs in the response payloads.
> I think a cleaner payload would be: ...
Unless I overlooked something the two example payloads are identical.
Regardless I agree that current JSON responses could be improved.
However in this particular case I am not sure whether "duplicities" are
an issue. The requested resources (services) just happened to have the
same parent. Also extracting these duplicities would mean that resources
listed under .../services and /services/{id} would be different, which
doesn't make sense as the former is just a list of all services
available after providing {id} as part of the url.
For my point of view a possible issue is the use of "id" attributes --
such attributes lack semantics and resource urls should be used instead
(as it allows transition between resources) [1]. Nevertheless there is
a plenty of APIs which are using ID attributes so I would consider this
to be of minor importance.
> When an organization is created, seems like organizationId is derived
from organizationName? What's the purpose of two fields uniquely
identifying an organization?
As you said, id is derived not copied -- some characters (such as
whitespace) are removed or replaced, thus the value of both fields can
be different. Name is meant for human users to read, id as a system
identifier.
>Can id, name, description attributes for service be renamed to
serviceId, serviceName, serviceDescription?
The requested resource is list of all services. I would kinda assume
that all attributes (unless something indicates otherwise) are
attributes of service object. So what would be the point of making the
attribute names more complicated? Again changing the attributes here
would require to mirror these changes to .../services/{id}
[1] Interesting presentation about REST api design. Unfortunately author
mixed in some Czech for no reason. The important stuff is in English
though.
https://speakerdeck.com/honzajavorek/rest-api-manual
Jakub.
On Thu, Jul 23, 2015 at 12:49 AM, Arun Gupta <arun.gupta(a)gmail.com
<mailto:arun.gupta@gmail.com>> wrote:
Moving to swagger will be good!
Can reasonable defaults be assumed? For example, not setting
endpointType currently sets it to null. It'll be convenient if it is
assumed it to be "rest". That is going to be the most common use case
anyway. How does the processing differ if the endpointType is set to
"soap"?
Another question ...
Accessing the list of services for an organization as:
curl --user admin:admin123! -H "Accept: application/json"
http://192.168.99.103:8082/apiman/organizations/everest/services
And the generated payload is:
[{"organizationId":"everest","organizationName":"everest","id":"catalog","name":"catalog","description":"catalog
service","createdOn":1437575090081},{"organizationId":"everest","organizationName":"everest","id":"order","name":"order","description":"order
service","createdOn":1437575867804},{"organizationId":"everest","organizationName":"everest","id":"user","name":"user","description":"user
service","createdOn":1437575649446}]
A few issues with this ...
- Why organizationId and organizationName are repeated with each
service? I think a cleaner payload would be:
[{"organizationId":"everest","organizationName":"everest","id":"catalog","name":"catalog","description":"catalog
service","createdOn":1437575090081},{"organizationId":"everest","organizationName":"everest","id":"order","name":"order","description":"order
service","createdOn":1437575867804},{"organizationId":"everest","organizationName":"everest","id":"user","name":"user","description":"user
service","createdOn":1437575649446}]
- When an organization is created, seems like organizationId is
derived from organizationName? What's the purpose of two fields
uniquely identifying an organization?
- Can id, name, description attributes for service be renamed to
serviceId, serviceName, serviceDescription?
Thanks,
Arun
On Wed, Jul 22, 2015 at 10:17 AM, Eric Wittmann
<eric.wittmann(a)redhat.com <mailto:eric.wittmann@redhat.com>> wrote:
> Fair questions all.
>
> - Everything is optional in the payload so that you can set
individual
> properties without sending *everything*
> - Only the 'gateways' has a default value - but it will default
only if
> there is a single gateway installed in apiman. if multiple
gateways have
> been configured, then you must pick one
>
> We'll likely be moving the API documentation over to swagger at
some point,
> replacing the miredot docs.
>
> -Eric
>
> On 7/22/2015 9:02 AM, Arun Gupta wrote:
>>
>> Here is the inbound payload for
>>
>>
http://localhost:8080/apiman/organizations/organizationId/services/servic...:
>>
>> {
>> endpointType:rest | soap
>> publicService:boolean
>> endpointProperties:{
>> string =>string
>> }
>> gateways:[
>> {
>> gatewayId:string
>> }
>> ]
>> plans:[
>> {
>> version:string
>> planId:string
>> }
>> ]
>> endpoint:string
>> }
>>
>> How do I know which parts of optional? Why "gateways" is an array
of
>> objects instead of array of string?
>>
>> Can there be defaults assumed if nothing is specified in the
payload?
>> For example, make the service public by default, use rest type,
assign
>> no plan, and have no endpoint properties.
>>
>> Created
https://issues.jboss.org/browse/APIMAN-559 for syntax
highlighter.
>>
>> Arun
>>
>> On Wed, Jul 22, 2015 at 8:33 AM, Eric Wittmann
<eric.wittmann(a)redhat.com <mailto:eric.wittmann@redhat.com>>
>> wrote:
>>>
>>> There is a feature request already in JIRA for allowing the
POST (create
>>> new
>>> service version) to include all of the same information that a
PUT would
>>> allow. A future version will allow this.
>>>
>>>
https://issues.jboss.org/browse/APIMAN-427
>>>
>>> Which optional parts are not indicated?
>>>
>>> -Eric
>>>
>>> On 7/21/2015 10:19 PM, Arun Gupta wrote:
>>>>
>>>>
>>>> POST on
>>>>
>>>>
http://localhost:8080/apiman/organizations/{organizationId}/services/{ser...
>>>> takes only:
>>>>
>>>> {
>>>> version:string
>>>> cloneVersion:string
>>>> clone:boolean
>>>> }
>>>>
>>>> and returns all the information about the created endpoint,
including
>>>> all the information that needs to be set later by calling PUT on
>>>>
>>>>
>>>>
http://localhost:8080/apiman/organizations/{organizationId}/services/{ser....
>>>>
>>>> Was there discussion about taking in the relevant information
as part
>>>> of original POST request?
>>>>
>>>> Also, optional JSON parts are not indicated in the payload.
How can
>>>> that be done?
>>>>
>>>> Can an appropriate syntax highlighter be used?
>>>>
>>>> Arun
>>>>
>>>> On Tue, Jul 21, 2015 at 6:11 PM, Eric Wittmann
>>>> <eric.wittmann(a)redhat.com
<mailto:eric.wittmann@redhat.com>>
>>>> wrote:
>>>>>
>>>>>
>>>>> It depends on a number of variables. But basically for each
service
>>>>> you
>>>>> want to create you'll need at least:
>>>>>
>>>>> 1) create the service
>>>>> 2) create version of the service
>>>>> 3) update version with the desired settings (endpoint info,
plans, etc)
>>>>> 4) publish version (pushes it to the gateway)
>>>>>
>>>>> That would be for a public service with no policies.
>>>>>
>>>>> If you want to configure it with policies, then there are
extra calls
>>>>> for
>>>>> that. If you wanted to add a service definition (swagger)
then there's
>>>>> a
>>>>> separate call for that. Etc.
>>>>>
>>>>> You'll of course need an organization to hold the
services.
You also
>>>>> need
>>>>> to create and lock plans if you intend to use multiple
plans. If you
>>>>> use
>>>>> plans (vs. public services) you'll need to create
applications and then
>>>>> contracts between the apps and the service(s). :)
>>>>>
>>>>> -Eric
>>>>>
>>>>> On 7/21/2015 5:06 PM, Arun Gupta wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> This is helpful!
>>>>>>
>>>>>> What is the minimum number of calls required for
registering
endpoints
>>>>>> for a few services?
>>>>>>
>>>>>> Arun
>>>>>>
>>>>>> On Tue, Jul 21, 2015 at 7:11 AM, Eric Wittmann
>>>>>> <eric.wittmann(a)redhat.com
<mailto:eric.wittmann@redhat.com>>
>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> There is a test plan in apiman that does all of this.
It
can be
>>>>>>> found
>>>>>>> here:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
https://github.com/apiman/apiman/blob/master/test/suite/src/main/resource...
>>>>>>>
>>>>>>> Each of the tests in the XML file is executed in
order,
starting with
>>>>>>> this
>>>>>>> one:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
https://github.com/apiman/apiman/blob/master/test/suite/src/main/resource...
>>>>>>>
>>>>>>> The format of each *.resttest file is:
>>>>>>>
>>>>>>> VERB /path/to/API/resource user/pass
>>>>>>> Request-Header-1: value
>>>>>>> Request-Header-2: value
>>>>>>>
>>>>>>> {
>>>>>>> "json" : "payload"
>>>>>>> }
>>>>>>> ----
>>>>>>> expectedResponseCode
>>>>>>> Expected-Response-Header-1: value
>>>>>>> Expected-Response-Header-2: value
>>>>>>>
>>>>>>> {
>>>>>>> "expected-json" :
"response-payload"
>>>>>>> }
>>>>>>>
>>>>>>> So basically you would be interested only in what is
above
the "----"
>>>>>>> separator.
>>>>>>>
>>>>>>> You are right - we should write a blog post or document
a
hello world
>>>>>>> use-case.
>>>>>>>
>>>>>>> -Eric
>>>>>>>
>>>>>>>
>>>>>>> On 7/20/2015 7:11 PM, Arun Gupta wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Is there a sample that shows the complete
creation/order of
>>>>>>>> organization, services, endpoints, etc using the
REST API?
>>>>>>>>
>>>>>>>> The documents at [1] are helpful but a Hello World
sample
would be
>>>>>>>> very useful. I started creating a sample at [2].
>>>>>>>>
>>>>>>>> [1]
http://www.apiman.io/latest/api-manager-restdocs.html
>>>>>>>> [2]
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
https://github.com/arun-gupta/microservices/blob/master/microservice/dock...
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> Arun
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
--
http://blog.arungupta.me
http://twitter.com/arungupta
_______________________________________________
Apiman-user mailing list
Apiman-user(a)lists.jboss.org <mailto:Apiman-user@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/apiman-user