I agree with Rich, although mentors need to be quite hands on at this
point. Do call upon them to conduct reviews and provide pointers to
appropriate supporting documents. The whole process is quite tortuous
the first few times. Once you get that first release out the door it
starts to get easier as you fine tune your processes.

With that in mind you need to start documenting the process early.

You might want to use a document such as  [1] as a starting point for
your project specific processes.

Ross

[1] http://wookie.apache.org/docs/developer/release-management.html
Ross Gardler (@rgardler)
Programme Leader (Open Development)
OpenDirective http://opendirective.com




On 18 April 2013 17:21, Rich Bowen <[email protected]> wrote:
> I submit that the very best way to learn about the right way to do a release 
> is to do a release, and then see what the Incubator mailing list tells us we 
> did wrong.
>
> I nominate Peter as release manager. :-)
>
> --Rich
>
> On Apr 18, 2013, at 11:54 AM, Peter Hartmann wrote:
>
>> As part of our graduation process from Incubator, we (Allura contributors) 
>> must demonstrate our ability to make an official release, in line with ASF 
>> guidelines. It seems that throughout last year all technical and legal 
>> obstacles have been removed and we're ready to start working on getting a 
>> release out. As per https://sourceforge.net/p/allura/tickets/3905/ we also 
>> want to make a release on PyPI and somehow synchronizing these two releases 
>> would be a sane thing to do.
>>
>> The most basic way we may do Incubator release is simply get a repository 
>> tarball on a public Apache server. This has the merits of being very 
>> straightforward and it seems to be most popular option among other projects.
>>
>> I've proposed on IRC that we may also release every module that forms Allura 
>> codebase as a separate tarball. The rationale behind this is that in process 
>> of making a release on PyPI, each module will have to be packaged separately 
>> anyways. PyPI/pip/setuptools also can work with tarballs and by keeping a 
>> single set of distributed archives we can avoid possible confusion on that 
>> part. Same tarballs would be uploaded to PyPI and Apache servers.
>>
>> It would be an unusal way to do things, but doesn't seem to conflict the 
>> guidelines: http://incubator.apache.org/guides/releasemanagement.html
>>
>> Release process can of course be modified later on, and the way we make 
>> releases is important enough to be discussed by whole community here. 
>> Anyone's input is welcome, so are any new proposals on the matter.
>
> --
> Rich Bowen
> [email protected]
> Shosholoza
>
>

Reply via email to