Hi,

this is a straighforward patch (and test) to remove the "gz" extension
in the XML output of a patch.

It also changes the behaviour of Darcs.Patch.patchname for consistency.
See my comment below about this.

Best regards,
Gabriel

4 patches for repository http://darcs.net/screened:

Thu Dec 16 16:46:47 CET 2010  Gabriel Kerneis <kerneis@pps.jussieu.fr>
  * Split makeFilename to get rid off the .gz suffix

Thu Dec 16 16:54:23 CET 2010  Gabriel Kerneis <kerneis@pps.jussieu.fr>
  * Remove the .gz suffix in Darcs.Patch.patchname
  
  This changes the behaviour of patchname in an **incompatible** way.
  However, I believe this is the right thing to do because:
  - although patchname is exported in Darcs.patch, it's behaviour is not
    documented,
  - patchname is not used anywhere in Darcs code,
  - removing ".gz" is consistent with darcs changes --xml (see issue1558).
  
  Note that it might be better to remove patchname completely if it turns
  out not to be useful anymore.
  

Thu Dec 16 16:54:28 CET 2010  Gabriel Kerneis <kerneis@pps.jussieu.fr>
  * resolve issue1558: remove ".gz" extension in xml output

Thu Dec 16 17:00:34 CET 2010  Gabriel Kerneis <kerneis@pps.jussieu.fr>
  * Test issue1558: xml output for patch hash shall not include "gz" extension

Attachments

• split-makefilename-to-get-rid-off-the-_gz-suffix.dpatch
• unnamed

These look good and are all fairly straightforward, so I don't have any 
detailed comments.

Basically we're getting rid of that pesky .gz suffix you see in darcs 
changes --xml patch names.

Some notes

1. It may be good to redo the haddock for makeFilename and 
makePatchname, perhaps moving it to makePatchname and tweaking a bit.
Maybe makeFilename could explain that it's just makePatchname plus gz 
and maybe say a word about why we want this (for old-fashioned repos?)

2. The patch 'Remove the .gz suffix in Darcs.Patch.patchname' depends on

  * patch481 disentangle Darcs.Patch.Named from Darcs.Patch.Viewing
  * patch481 break out Summary code into separate module
  * patch482 get rid of Split
  * patch483 drop identity member of Invert

I think I agree with Gabriel's justification for the patch and also that 
patchname should probably just be removed (we already have a function 
that just extracts PatchInfo).  But perhaps Ganesh will have some API 
design wisdom that says otherwise.

I've pushed what I can in the meantime

Pushing last patch 'Remove the .gz suffix in Darcs.Patch.patchname' to
screened and darcs.net now that deps are in.