db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel John Debrunner <...@apache.org>
Subject Re: Beta Announcement and encouraging user testing
Date Thu, 10 Aug 2006 17:26:05 GMT
Rick Hillegas wrote:

> Thanks to everyone for thinking through the issues here. This has been
> very helpful. I don't understand the practical advantages of a copy
> versus a branch, particularly since I will have to create a branch
> eventually. I could use some more education here. What would be the
> objections to the following scheme? A big advantage to this scheme is
> that it's something I can wrap my mind around:
> 
> 1) I create a 10.2 branch.
> 
> 2) I turn on the beta bit in that branch.
> 
> 3) I generate the beta candidate from that branch.
> 
> 4) In the meantime, bug fixes accumulate in the trunk.
> 
> 5) At some point I mega-merge the trunk into the 10.2 branch based on a
> triggering event:
> 
>  a) triggered by the need to generate a new release candidate
> 
>  b) triggered by the checkin of destabilizing work such as a partial
> feature
> 
> 6) After the mega-merge, fixes must be applied to both the trunk and the
> 10.2 branch.
> 
> Thanks again for helping me work through these issues.

I think the point to the beta branch and a real 10.2 branch later is:

  1) branches (copies) are cheap in svn, so no major cost there.

  2) There is no "mega-merge" required.

  3) history is cleaer, easy to see 10.2 branch created from trunk @
503134. The mechanism you propose has the history being:
      10.2branch created from trunk @ 489243 (made up numbers btw)
      Big merge of stuff from trunk between 489244-503134
      (was this done right?, was a file missed?, all possibilities).

Dan.



Mime
View raw message