[cvsnt] Re: Branch merging - this seems wrong...

Tony Eva teva at Airspan.com
Thu Jun 8 09:31:15 BST 2006


Gerhard Fiedler wrote:
> Why does there need to be a "better" solution -- does the one 
> given not cover 100% of what you wanted?

It does cover 100% what I need and it is what we will be doing with
CVSNT.  All I meant was that it means that the tool is forcing a process
requirement - that development on the B *must* finish with a final merge
from A to B to bring B up to date, otherwise the merge/copy back to A
will produce the wrong result.  However since the final merge/test is an
essential part of good design practice anyway, it is not a bad thing and
we will in all likelihood use a script to automate the 'close branch B
and merge back to A' procedure, with appropriate checking to ensure that
the branch is fully up to date first.

I am not intending to continue the discussion -- now I have a workaround
I can use -- though I remain very firmly of the opinion that CVSNT does
the wrong thing in the 'back and forth' merge situation.  The opposing
views on this seem solidly entrenched, though, and I don't think any
amount of discussion will alter that.  However, thanks to all who
participated for keeping it civil and informative :-)

> I don't see why you would need temp files. [...]

Yes, the 'update -jA -jB' command appears to do what's needed to achieve
the copy.  Thanks.

-- 
Tony



More information about the cvsnt mailing list