Re: Release Review (was: Incubator September Report)

Posted by David E Jones-2 on
URL: http://ofbiz.116.s1.nabble.com/Incubator-September-Report-tp171490p171515.html


On Sep 9, 2006, at 9:56 AM, Yoav Shapira wrote:

> On 9/9/06, David E Jones <[hidden email]> wrote:
>>
>> On this topic... I've been looking at the release stuff and decided
>> to just go for it...
>
> In the future a heads-up, at least to the PMC but preferably to the
> public ofbiz-dev list, would be appreciated ;)  Such a heads-up lets
> other committers make sure they've committed what they're done with
> and reverted experiments that could negatively impact a release.
>
>> For this initial RC1 for version 4.0.0 I decided to go with the tag
>> route and we'll do a branch later when we're ready to do a final
>> release.
>
> +1.
>
>> The "tag" in this case is just an svn "soft" tag, ie the
>> implicit tag from the revision number (see the REVISION) file in the
>> archive. A real tag in SVN is pretty much the same as a branch and I
>> think is a little heavier than what we need for this.
>
> OK, but just so you know SVN tags are cheap and light: it's a single
> (atomic) copy operation, that's it.  A real SVN tag provides people
> with easy ways (including easily scriptable ways) to recreate past
> builds without needing to find the corresponding text file and read
> it.

These are good points. For this "release" the intent, as I understand  
it, is to review the release process and all of the artifacts that  
need to be in, and with, the release files. That is why I tried to  
focus on those things below in the list of items to review. If there  
are things partially completed (well, I know there are many... but  
that's another problem/topic altogether...) as long as they are not  
show-stoppers it should be okay for this pass.

Because we are not doing a branch, which is planned to be our main  
quality control mechanism for releases, I'm hoping no one will  
actually _use_ this "release" as one would normally use and rely on a  
release to be better than the latest trunk revision.

I'm also guessing that given that this is our first time going  
through this we'll almost certainly mess it up, and so we may have an  
RC2 before it gets past the OFBiz PPMC.

>> 1. test un-archive and run as-is (should only require JDK installed)
>> 2. identify and show-stopper bugs that we should fix before a real RC
>> 3. check all ASF required files (NOTICE, LICENSE, KEYS, etc) and  
>> headers
>> 4. check PGP and MD5 files
>> 5. anything else?
>
> Are there any automated unit tests done as part of the release
> creation?  If so, the person cutting the release should also let us
> know whether they all passed (hopefully), or if any failed, which
> ones.

Yes, there are some unit tests. With the way it is setup in OFBiz  
(for the limited automated tests that there are) you just run "ant  
run-tests". This should be done on the same revision and after the  
"ant run-install" but the tests are just run for testing and the  
result after running the tests doesn't go into the release, if that  
makes sense. It may be that in some cases the tests don't clean up  
100% after themselves, and to make them easier to develop we don't  
want to force this.

Anyway, for this release all of the unit tests were successful. Of  
course, there are only 2 components with tests with 1 + 23 = 24 total  
tests. These are just framework level things right now. This is one  
of the directions of the project, but hasn't really received a lot of  
attention or demand (or in other words: resources...).

>> The files are posted for initial review as attachments on the
>> docs.ofbiz.org site to this page:
>>
>> http://docs.ofbiz.org/x/wAE
>
> Since this is a public page, please add a clear disclaimer to let
> people know this is not an official release and hasn't been vetted by
> the PMC.

Great, good point. I added a comment to that effect on the wiki page.

One thing I was just thinking of about (not necessarily related to  
the other comments in this email) that page on Confluence is that  
only the committers can post to that area, so there is an added level  
of confidence that the files posted come where the pages say they  
come from.

Thanks for your feedback.

-David