[Gnu-arch-users] Cascading patchlogs with star-merge

[Russell]

Hi,

Distributed development with multiple archives, using star-merge to sync
the working copies and archives looks very attractive, and seems to work
very well.  However I've run into a situation where I can generate
endless patch logs from one change.  It goes like this:

  Commit patch-N in archive 1
  star-merge from archive 1 into archive 2, then commit, creating patch-M there
  star-merge from archive 2 into archive 1, commit creating patch-N+1 there
  star-merge from archive 1 into archive 2, creating patch-M+1
  ... etc

This can go on forever, and the only things getting added to the working
copies each time are patch logs.  I was looking to automate the sharing
of patches like this, but found that it never stopped.

"Don't do that!", I suppose.  Is this a communication thing, where it's
up to each person to know that something has changed, and then only do
merging when it's known to be necessary?  I had expected, from reading
the tutorial, that arch would stop once all the source changes had made
their way into each archive.

How do people handle this in practice?




-- 
Russell Steicke

-- Fortune says:
"It's what you learn after you know it all that counts."
-- John Wooden