[darcs-users] darcs annotate format obscures the code

Ivan Stankovic pokemon at fly.srk.fer.hr
Sun Mar 20 14:44:15 UTC 2005


On Sun, Mar 20, 2005 at 01:58:50PM +0000, Ralph Corderoy wrote:
> Hi Jamie,
> 
> > On Sun, Mar 20, 2005 at 12:33:21PM +0000, Ralph Corderoy wrote:
> > > The first line that doesn't match /^[ 1-9]/ will always be the end
> > > of file line so something more simple like `.' as used by ed(1),
> > > SMTP, etc., may be less noisy.  Alternatively, allow the explanation
> > > of the numbers to be in a separate file.
> > 
> > I'd rather see the numbers in the output of darcs changes. That way I
> > can run annotate in one window and changes in another. Modification of
> > the repo shouldn't be a problem because a) presumably this is usually
> > being done in a private copy
> 
> Usually isn't good enough to avoid misleading output to the user.
> 
> > and b) any new patches will just get higher numbers.
> 
> I don't think that's true.  They're date ordered and the date is when
> the patch was recording, not when it was added to this repo, so
> introducing a patch written six months ago will bump lots of numbers on
> by one.

So how about adding a patch attribute like 'date of addition to the
repository'? This attrubute would differ from record-date only if the
user pulled the patch from another repository. With this in place, you
could make 'darcs changes' and 'darcs annotate' build a sorted list of
patches and number each patch accordingly. This would work even if you
used unpull/unrecord since the list would be created on-the-fly, rather
than stored somewhere (also, you'd have to be careful and not assume
that patch #116 would be the same as patch #116 three weeks later; in these
cases you can always use full patch names).

Personally, I'd love to be able to refer to a patch by a simple number
instead of a regexp.
-- 
Ivan Stankovic, pokemon at fly.srk.fer.hr




More information about the darcs-users mailing list