On Thu, Apr 6, 2017 at 4:41 PM, Fabrice Bacchella <fabrice.bacchella@orange.fr> wrote:

Le 6 avr. 2017 à 15:32, Yaniv Kaul <ykaul@redhat.com> a écrit :



On Thu, Apr 6, 2017 at 3:58 PM, Fabrice Bacchella <fabrice.bacchella@orange.fr> wrote:
Yes I'm starting to understand that thinking about migrating code is pointless.

The old skd3 code is just good to be thrown away. There is no hope thinking about "migrating code". And as it's just a thin layer around REST calls, it's up to us to try to make something usable around that. So I expect a lot of sweat and tears to adapt my existing code.

Well, yes and no. Yes, it's not smooth, but once you 'get' the idea behind the v4 API philosophy, it's quite easy to write to (at least in Python).

Easy to write code that a well though sdk should have provided.

I'd like to believe our v4 SDKs are. They fix several inconsistencies we've had with v3.
  

Note that right now you can mix between v3 and v4, so you can migrate slowly, function by function.

That's a possible but almost as complicated as rewrite everything in my case.

Perhaps in your case. Here[1] is an example of the ovirt system tests, which were only partially converted (work in progress...) to v4 API.

HTH,
Y.

[1] https://gerrit.ovirt.org/gitweb?p=ovirt-system-tests.git;a=blob;f=basic-suite-master/test-scenarios/002_bootstrap.py;h=c0e71df23f17335d59d9d37fcd4c833be6372c69;hb=HEAD#l109




Another option that you can consider, if you are re-writing, is automation via Ansible. 

A lot of people don't use ansible or use concurrent tools. So no that's not an option.