Dear Darcsers

while I'm giving you trouble, I was surprised when 

	darcs unrecord -p 'f'

showed me patches *not* beginning with "f".  

Incidentally, darcs check is happy; and I'm using version 1.0.8

Simon

sh-2.04$ darcs unrecord -p 'f'

Fri Sep 29 16:14:15 GMT Daylight Time 2006  simonpj@microsoft.com
  * f
Shall I unrecord this patch? (1/?)  [ynWvpxqadjkc], or ? for help: y

Fri Sep 29 15:39:10 GMT Daylight Time 2006  simonpj@microsoft.com
  * Global renamings in HsSyn
Shall I unrecord this patch? (2/?)  [ynWvpxqadjkc], or ? for help: n

Fri Sep 29 14:35:49 GMT Daylight Time 2006  simonpj@microsoft.com
  * Improve pretty-printing of Core
Shall I unrecord this patch? (3/?)  [ynWvpxqadjkc], or ? for help: ?

How to use unrecord...
y: unrecord this patch
n: don't unrecord it
w: wait and decide later, defaulting to no

simonpj wrote:
> Dear Darcsers
> 
> while I'm giving you trouble, I was surprised when 
> 
> 	darcs unrecord -p 'f'
> 
> showed me patches *not* beginning with "f".  

I think there's two things going on here:
1. 'f' matches all patches with names containing f, not just names 
beginning with f
2. darcs will also ask you about any patches that depend on a matched patch.

Edwin

On Fri, Sep 29, 2006 at 03:59:16PM +0000, Edwin wrote:
> simonpj wrote:
> > 	darcs unrecord -p 'f'
> > 
> > showed me patches *not* beginning with "f".  
> 
> 1. 'f' matches all patches with names containing f, not just names 
> beginning with f

To be precise, darcs matches the regexp "f" against the names of
all patch after the last inventory break.

  darcs unrecord -p '^f'

will choose patches beginning with f.

  darcs unrecord --from-<something>=<regexp> -p <regexp>

can go further back than the last inventory break.

Users are not supposed to know what "last inventory break" is.
It is for performance reasons darcs "have" to do this cut-off.
It is not likely that one wants to remove patches that are
taged, and the inventory break is always a tag.

Regexps will enter into most darcs users' lives at some point
though.