[darcs-users] renaming the rerecord command

[Andrew Pimlott]

On Tue, Oct 19, 2004 at 05:42:57AM -0400, David Roundy wrote:
> Hmmmm.  I think amend sounds reasonable.  I do like the fact that rerecord
> has the word "record" in it, but this is a very easy mistake to make, and I
> certainly find myself telling people to unrecord a patch and then rerecord
> it, and I don't mean to use "rerecord".  Since it's quite a new command,
> I'm amenable to changing it's name even this close to 1.0.0, but would like
> a bit more discussion, in case other people find "amend" confusing, or like
> "rerecord".

This is a knotty issue.  I find "rerecord" very natural for its current
function, yet it undeniably also sugests the opposite of unrecord.  As
Dilbert's boss would say, "let's do both".  Here's my idea:

Short term:
- Simply add a line of output when you run rerecord that says, "warning:
  rerecord is not the opposite of unrecord; try darcs record", and
  change the prompt to "Shall I rerecord this currently recorded
  patch?".  This is kind of lame, but should avoid disasters, and will
  tide us over until...

Long term:
- Store the unrecorded patch, with its context so that it can be
  reapplied, and have rerecord ask about it when rerecord is run.  So it
  starts out by asking

    UNRECORDED PATCH
    Mon Oct 18 18:01:17 PDT 2004  andrew at pimlott.net
      * a
    Shall I rerecord this unrecorded patch? [yNvq?] n

    Tue Oct 19 13:28:51 PDT 2004  andrew at pimlott.net
      * b
    Shall I rerecord this currently recorded patch? [yNvq?] n

This way, whether people expect rerecord to be the opposite of unrecord,
or do what it does today, everyone will find what he wants and nobody
will get tripped.

But maybe this is too much work.  I'm sure I could get used to "darcs
amend" or "darcs editpatch" (my suggestion).  However, I think
"rerecord" is more discoverable than the other suggestions.

Andrew