Simplify init-repo

[Matthew D. Fuller]

On Tue, Feb 19, 2008 at 11:13:47AM +1100 I heard the voice of
Robert Collins, and lo! it spake thus:
> On Tue, 2008-02-19 at 10:33 +1100, Andrew Bennetts wrote:
> > The only bit I'm hesitant about is hiding --knit, [...]
> 
> IIRC Alexander prefers knit for windows 98 over dirstate due to
> different os locking usage.

Er.  Well, this is a little off the point I was aiming at...  this is
more along the lines of "clean up format list".  While certainly
something to look at, I was talking more about meaninglessness in
init-repo specifically.  For instance:

- init-repo --weave is nonsensical.  Even worse (now trying it),
  trying it actually *DOES* something, but it doesn't init a shared
  repo.

- knit/dirstate/dirstate-tags don't have a meaningful difference in
  this case, since the repo format is the same across them all.  And
  the repo doesn't influence them at all either, so if you init-repo
  --knit, then init a branch within it, that branch will be
  dirstate-tags with current versions (or do it the other way around
  with init-repo --dirstate-tags on versions prior to branch6 being
  default).


We're using all the same args for init and init-repo, which saves some
work certainly, but it leads to weirdness like the above.  That
entries also show up for deprecated formats is probably a different
sort of bug.


-- 
Matthew Fuller     (MF4839)   |  fullermd at over-yonder.net
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.