On 2015-06-15 14:49 Ignasi Barrera wrote:
Would it be possible to set a sort of "final date" for such refactoring to be completed, after that, if nothing will have changed in terms of PR [2][3]
progress, the provider will be anyway released with 1.9.1?

Yes. I'd say so. I think a reasonable plan would be to release 1.9.1
with or without the refactoring (we'll see if it is on time) and
promote in master what we release. If we don't release the refactoring
we can promote it directly, as it would be pretty much the same than
the provider released in 1.9.0.

Fine.

The reason to promote it in master and not in 1.9.1 is to avoid a
groupId name change in a bugfix release. Does this make sense?

I haven't thought about this: sure, it makes sense.

Any timing indications for 1.9.1 or 2.0.0 (days / weeks / months)?

Regards.

On 15 June 2015 at 14:40, Francesco Chicchiriccò <ilgro...@apache.org> wrote:
On 2015-06-15 12:34 Ignasi Barrera wrote:

Hi!


Hi Ignasi,

It's been a while since our last release and I'd like to start
planning jclouds 1.9.1. Many bugfixes have been introduced and there
are several providers that have been improved a lot, including GCE,
DigitalOcean, Azure and ProfitBricks.

Before talking about dates, I'd like to share the important topics
(IMO) that should be finished for 1.9.1:

* Merge the ProfitBrixks ComputeService integration [1].
* Complete the Azure Compute refactor [2, 3]. There is an ongoing
effort to improve how virtual machines are created but it is moving
slowly. @andreaturli, @ilgrosso, once the PRs are up to date, would
you be able to help reviewing them?


As you're saying, it seems there is a considerable latency there: I thought we could promote the AzureCompute provider with 1.9.1 without the above
mentioned refactoring being finalized.
Would it be possible to set a sort of "final date" for such refactoring to be completed, after that, if nothing will have changed in terms of PR [2][3]
progress, the provider will be anyway released with 1.9.1?

* Merge the DigitalOcean v2 provider [4, 5]. There is still no PR, for
it but the provider is ready to go.


Considering that we are in time to finish this, we'll be in a good
position to promote the following providers (only in master, for
2.0.0), as they'll have been out there for a while with the live tests
passing:

* GCE. I plan to promote it this week.
* Azure. It's been out since 1.9.0 and will remain in labs after the
compute service refactoring. Should be ready to be promoted for 2.0.0.


See my comments above.


* Profitbricks. Same thing. @devcsrj is doing a fantastic job keeping
it up to date and having it working for in 1.9.1 with all live tests
passing gives us and our users enough time to test and get used to it.
I see no reason to keep it in labs.
* DigitalOcean v2. Same thing. The v1 API has been stable for ages but will be shutdown on November 2015. We shouldn't promote it but promote
the v2 for 2.0.0. It's also stable and the interface hasn't changed.
The internals are different (different api calls) but the behavior,
the results, are the same, so users using the ComputeService shouldn't
experience any change.



Please, feel free to give your feedback, add what you would want to
see in 1.9.1. Your opinions are very welcome!


Ignasi




[1] https://github.com/jclouds/jclouds-labs/pull/145
[2] https://github.com/jclouds/jclouds-labs/pull/175
[3] https://github.com/jclouds/jclouds-labs/pull/157
[4]
https://github.com/ccustine/jclouds-labs/tree/features/digitalocean2final
[5] https://github.com/nacx/jclouds-labs/tree/do-pagination
--
Francesco Chicchiriccò

Tirasa - Open Source Excellence
http://www.tirasa.net/

Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC
http://people.apache.org/~ilgrosso/

Reply via email to