Next release

[Diogo Baeder]

Hi guys!

So, I've bumped provy's version and release to 0.6.1-dev, and it's changed in RTD already. The idea is to release a 0.6.1 version soon, because we had some API changes that I'd like to have released and keep in sync with the docs.

So, here are the next steps I'll go through, in the next few days:

• Wait for some more changes from Jacek to merge (fixes only, no new features);
• Bump to 0.6.1 (no "-dev");
• Create a 0.6 branch, to keep evolving with bugfixes only;
• Bump to 0.7.0-dev, so that we start the next changes which affect our API (and hence generate different documentation).

So, before 0.6.1 is out, I won't merge changes which affect the API; Only after 0.6.1 is out, and we have a 0.6 branch, I'll go on with these merges, so that they don't affect the old documentation ("latest" will keep changing).

Please keep in mind that these procedures are to help us providing a reliable documentation to our users.

Cheers!

__________________________
Diogo Baeder
http://diogobaeder.com.br

[Bernardo Heynemann]

Hey man,

Just wanted to drop by to say you are doing an incredible, amazing, fantastic job.

Regarding the 0.6 branch, I'd rather have tags. This way if you do need to fix something in the 0.6.x releases, all you need to do is:

git checkout 0.6.x (where x is the latest tag in this series)

git checkout -b 0.6

...make changes and commit

git push

git tag 0.6.x+1

git push --tags

Now we do have a 0.6 branch. I think it can be safely removed without the tag being affected, but I have to test it. Even if we can't I think it's better to have a "lazy" bugfix branch than having one per minor.

Just nitpicking anyways. If you decide to go with the release branches I'm with you :)

Btw, I will pick some tickets to implement with tests.

One thing that would be really cool but we have to think about is an integration suite that can be run against vagrant VMs.

It will take a long time to run, but it should just be a checkpoint against releases, not something we'd run in the CI.

What do you guys think?

Cheers,

Bernardo Heynemann

Bernardo Heynemann
Developer @ globo.com

--
You received this message because you are subscribed to the Google Groups "provy" group.
To unsubscribe from this group and stop receiving emails from it, send an email to provy+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Diogo Baeder]

Well, I think that might work well with RTD too, I'm with you on it! :-)

OK, so, bugfix branches only when the bug actually appears.

I agree with the integration tests before releases, maybe even an end-to-end one, with as much features as possible that we can use. If we only had a server to run it, though, maybe we could even have it run nightly... do you know if some company (not necessarily globo.com) would be willing to donate a VM for us, for this? If so, I can install Jenkins (if the Skink project is not ready yet - btw, haven't had time to help you with it yet, sorry) so that we can have these nightly builds.

Glad to hear from you again! :-)

Cheers,

__________________________
Diogo Baeder
http://diogobaeder.com.br

[Fernando]

Running integration tests before releases would be great, running every day would be awesome =)

How do you plan to verify it everything is running fine after provisioning? Maybe monitoring tools plus some benchmark tool (as apache bench) to generate some requests for web roles.

[]s

http://about.me/fernandogrd

[Diogo Baeder]

You mean, after running these tests?

__________________________
Diogo Baeder
http://diogobaeder.com.br

[Fernando]

On 3 February 2013 00:21, Diogo Baeder <diogo...@gmail.com> wrote:

You mean, after running these tests?

Yes, but let's see if I understood it right. Will these integration tests be basically provyfiles? By saying 'integration suit', this mean just running provyfiles and verifying for provy errors on output or more than that?

--
http://about.me/fernandogrd

[Diogo Baeder]

Sort of. Not errors in the output, but checking that the final provisioning system serves certain requirements, like a web app running correctly (responding as expected). It would be interesting to do this with all supported operating systems, if possible.

But I don't see a reason to run benchmarks over it.

Cheers!

__________________________
Diogo Baeder
http://diogobaeder.com.br

[Fernando]

On 3 February 2013 02:36, Diogo Baeder <diogo...@gmail.com> wrote:

Sort of. Not errors in the output, but checking that the final provisioning system serves certain requirements, like a web app running correctly (responding as expected). It would be interesting to do this with all supported operating systems, if possible.

But I don't see a reason to run benchmarks over it.

 I mean not to run benchmarks, but to use benchmark tools, for example, to generate requests, because they are good at it =).

--
http://about.me/fernandogrd

[Diogo Baeder]

No need for it. We can just use "requests" and we're good. ;-)

__________________________
Diogo Baeder
http://diogobaeder.com.br