[Gnu-arch-users] more on the merge-fest

[Tom Lord]

I've now merged and tweaked (as per discussion):

~ per-directory regexps (.arch-inventory)
 (thanks robin)

~ multi-line regexp specs 
  (thanks walter)

And, overall: thanks rob.  I think that this integration-branch
experiment is a smashing success so far.    Specifically:

~ my workflow was sped up

  I didn't have to update lots of mirrors before the merge-fest,
  just one.   Even if modem-time was the only benefit, it would
  be a big one.   But it isn't the only benefit.   I didn't
  have to fuss with the horrible bug-tracker during merges
  and, what's more:


~ review was simplified

  Rob's integration branch isn't the platonic ideal of 
  clean isolated changesets (for which we'd want multiple
  integration branches anyway) but it is "close enough for
  government work".

  I've been cherry-picking, with the ultimate aim of being 
  fully up-to-date with it.   When I need to make tweaks to 
  a patch, I've been trying to remember to do that in separate
  changesets so that it will work with --skip-present.

  In one or two cases, the "patch dependency problem" showed up
  when I cherry-picked in the wrong order -- but `undo' made this
  easy recover from and then go back and do it in a better order
  later.  (Thanks alexander).


~ confidence was greatly increased

  What I got from rob was code that passed the test suite and
  that at least some bleeding-edgers had been using.   This 
  made it much easier to tweak the code to fix the glitches.
  For example, when removing the ".arch-inventory has a names
  tag" functionality the process was:

        1) fix the code
        2) verify that it broke the test suite
        3) study and fix the test suite

-t