openstack team mailing list archive

[George Reese <george.reese@xxxxxxxxxxxxx>]

On Aug 9, 2012, at 8:14 PM, Doug Davis <dug@xxxxxxxxxx> wrote:

> 
> Situations like this are always interesting to watch.  :-) 
> 
> On the one hand its open-source, so if you care about something then put up the resources to make it happen. 

This attitude always bothers me. This isn't some Open Source labor of love. It's a commercial collaboration in which many of the contributors have a significant economic interest. 

To be more blunt: if I'm writing code, it's for enStratus.


> On the other hand, that doesn't mean that as a developer you get to ignore the bigger picture and only do 1/2 of the work because you don't care about the other 1/2. 
> 
> Overall, I tend to agree with the attitude that as long as XML is officially supported then all code changes need to make sure they run through both the JSON and XML codepaths. And if this means twice the testcases then so be it.  People committing code shouldn't have a choice in this - its either you do the full job or your code is rejected. 
> 

+100000

> Having said that, it is a valid question to ask whether we want to continue to support both JSON and XML going forward.  But, until that decision is formally made letting 1/2 of the APIs atrophy makes the entire community look bad and therefore should not be allowed to happen.   
> 

Actually, it's not a valid question. That ship has sailed. XML is part of the spec, and we're talking about Folsom not Bexar. The API is the heart of the ecosystem. You never, ever deprecate code unless there's some strong compelling reason like a significant security flaw that can be addressed only through incompatibility.

> My vote: from now on don't let any code change in unless if works for both.  I suspect we'll either see the XML side come up to speed really quickly or it'll force an ugly vote.  But either way, this needs to be resolved before the next release. 
> 
> thanks
> -Doug
> ________________________________________________________
> STSM |  Standards Architect  |  IBM Software Group
> (919) 254-6905  |  IBM 444-6905  |  dug@xxxxxxxxxx
> The more I'm around some people, the more I like my dog. 
> 
> 
> George Reese <george.reese@xxxxxxxxxxxxx>
> 08/09/2012 07:02 PM
> Please respond to
> OpenStack Development Mailing List <openstack-dev@xxxxxxxxxxxxxxxxxxx>
> 
> To
> OpenStack Development Mailing List <openstack-dev@xxxxxxxxxxxxxxxxxxx>
> cc
> "openstack@xxxxxxxxxxxxxxxxxxx \(openstack@xxxxxxxxxxxxxxxxxxx\)" <openstack@xxxxxxxxxxxxxxxxxxx>
> Subject
> Re: [openstack-dev] [nova] Call for Help -- OpenStack API XML        Support
> 
> 
> 
> 
> 
> And this is why I go off on the developer-oriented mentality of the OpenStack community. 
> 
> The fact that there is no one in the OpenStack developer community writing XML stuff is not a reflection of the fact that there's no huge desire for XML. 
> 
> It's in the spec for a reason: BECAUSE ENTERPRISES USE XML HEAVILY 
> 
> OpenStack developers aren't that audience. They use JSON. 
> 
> That the project can get to this point and not have tests for these things shows a flaw in the development processes, not some grand illustration of supply and demand. 
> 
> Do I really have to point out that if the spec calls for JSON and XML, you should bloody well write integration tests to check for JSON and XML? 
> 
> You don't write whatever happens to please you. 
> 
> You know how I know all of this? I have an API that supports both XML and JSON. I personally prefer JSON. Most of my friends and colleagues prefer and use JSON. 
> 
> Most of my customers use XML. 
> 
> Thank $deity I actually write unit tests for each format. 
> 
> -George 
> 
> File under: 
> -> statistics 101 
> -> software development 101 
> 
> On Aug 9, 2012, at 5:52 PM, Vishvananda Ishaya <vishvananda@xxxxxxxxx> wrote: 
> 
> 
> On Aug 9, 2012, at 3:32 PM, George Reese <george.reese@xxxxxxxxxxxxx> wrote: 
> 
> Why aren't the integration tests both XML and JSON? 
> 
> The simple answer is that no one has taken the time to write them. Our devstack exercises use the python client bindings. Tempest has json clients but no xml clients[1]. I think this demonstrates that there just isn't a huge desire for xml. Users that I have chatted with just seem to care that the api works and that they they have good bindings. 
> 
> I am definitely willing to be proven wrong on this point, but I'm secretly hoping everyone agrees with me. It is a lot of work to maintain three APIs (we are still maintaining EC2 as well) and keep them all functioning well, so if people are happy without OpenStack XML I would be perfectly content to deprecate it. 
> 
> Vish 
> 
> [1] https://github.com/openstack/tempest/tree/master/tempest/services/nova/xml 
> 
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev@xxxxxxxxxxxxxxxxxxx
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev 
> 
> -- 
> George Reese (george.reese@xxxxxxxxxxxxx)
> t: @GeorgeReese               m: +1(207)956-0217               Skype: nspollution
> cal: http://tungle.me/GeorgeReese   
> 
> 
> 
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev@xxxxxxxxxxxxxxxxxxx
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~openstack
> Post to     : openstack@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~openstack
> More help   : https://help.launchpad.net/ListHelp

--
George Reese - Chief Technology Officer, enStratus
e: george.reese@xxxxxxxxxxxxx    Skype: nspollution    t: @GeorgeReese    p: +1.207.956.0217
enStratus: Enterprise Cloud Management - @enStratus - http://www.enstratus.com
To schedule a meeting with me: http://tungle.me/GeorgeReese