Andrea Aime wrote:
> Jody Garnett ha scritto:
> ...
>> Okay that is probably the step I was missing; with that in mind can we 
>> make the decision to keep the api we publish here stable for GeoServer 
>> 2.0 - even if that means we present new capabilities in GeoServer 2.0 
>> in a slightly screwed form (ie we may not get away with pure 
>> reflection?).
> 
> For 2.0 I believe so. But I don't have the crystal ball.
> If some significant contribution comes in, or some significant new
> functionality is sponsored, we may have to change the api.
> 
> As far as I remember in the REST world one tends to "version" the
> api and try to keep at least one old version around, not sure what
> Justin's plan are on this one thought.
The api is going to undoubtedly change as we continue to change the 
configuration. I have tried to hit a middle ground with the current rest 
api that hits some of the concepts that we anticipate, 1) it groups by 
workspace, not namespace, and 2) makes layer a separate entity when in 
reality it is not, etc.. Versioning the api makes a lot of sense to me. 
And when it does come to change I don't anticipate any issues with going 
though a normal deprecation cycle first.
> 
>>     To handle the kind of structural changes you seem demanding for, we
>>     need to work on trunk and change the catalog api accordingly, and we
>>     need some funding to make the changes, or someone that wants to
>>     spend his time working on it.
>>
>>
>> I was not aware this was pure reflection. 
> 
> It was figured speech, I was not implying the usage of Java reflection
> (thought it's actually used in the classes, but don't exactly know to
> which extent).
It uses reflection for serializing and deserializing catalog objects. 
The rationale was 1) we already have a bunch of utilities that help us 
do reflection and 2) we already had the xstream encoder/decoder which 
works reflectively so using it for rest was a pretty big win to support 
XML and JSON out of the box.
> 
>>     The topic of GSIP 33 is "is this REST api good to deal with the
>>     GeoServer catalog we have today?", what you're trying to discuss is
>>     "is the catalog API good for the
>>     next 5 years?" which is definitely a interesting topic, but a
>>     different one.
>>
>>
>> I was trying to hit the middle ground of; how do we intergrate a few 
>> of tomorrows features into the api being published today. I tried to 
>> make some suggestions that would agree with the plan out outlined in 
>> the documentation.
> 
> What makes people react negatively to your suggestion is that they
> come with request of more work and no contribution on your part.
> The idea is welcomed, but the way you're expressing it seems like
> twisting people arm into doing more work than they have resources for...
> not surprisingly you get negative reactions.
I think this may be a mis understanding of what feedback to a proposal 
should be. I think most people want feedback to be concise and something 
they can readily factor into a proposal, where as often your feedback is 
thinking much longer term and at times off topic.
> 
>>
>>     Let's talk about it, but in a different thread, and then let's put
>>     the result, along with an estimation, on the roadmap ideas, and see
>>     if anyone is up to funding it.
>>
>>
>> I do not think we need to go that far; I just wanted to account for 
>> how future work would be integrated into the api as published. I do 
>> not want to publish an API we know is going to be broken in the next 
>> year.
> 
> Then let's stop doing anything in GeoServer. The current development
> model does not allow to account for such long term concerns.
> Release early and often does not work well with "let's account for next
> 5 years". When we're lucky it has a glimpse of what's coming in the
> next year.
> 
> Just look, for example, at the roadmap ideas page. It has topic to cover
> one/two years of development. Yet none of them has direct sponsoring
> so far, so I cannot say whether they will be development in six
> months, or five years.
> 
>> Or if we do I would like to know that - and be up front with script 
>> writers that they will need to adjust their scripts to work with 
>> GeoServer 2.0.
> 
> That is certainly a concern I share. This is becoming published
> api, meaning that if we change it, we need to provide a migration
> path. Possibly going thru intermediate deprecations and replacements.
+1
> 
> Cheers
> Andrea
> 


-- 
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.

------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Geoserver-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

Reply via email to