This is interesting.  On two different Windows computers of my own,  
plus Arno's Windows computer, several different repos have turned out  
to be inconsistent once we upgraded to darcs v1.0.9.

Is there any known issue that would cause this?  I don't remember  
exactly what version of darcs I last used on these repos.  Probably a  
1.0.9 pre-release.  Arno was using 0.9.19.  :-)

I can give you copies of my inconsistent repos.

Regards,

Zooko

On Sat, Aug 11, 2007 at 17:56:46 -0000, Zooko wrote:
> This is interesting.  On two different Windows computers of my own,  
> plus Arno's Windows computer, several different repos have turned out  
> to be inconsistent once we upgraded to darcs v1.0.9.

Could you clarify this, please?  Do you mean that darcs check in 1.0.9
was able to detect inconsistencies that previous darcs versions did not?
Or do you mean that 1.0.9 introduced some new corruption?

> Is there any known issue that would cause this?  I don't remember  
> exactly what version of darcs I last used on these repos.  Probably a  
> 1.0.9 pre-release.

> Arno was using 0.9.19.  :-)

I thought in issue517, he said he was using 1.0.9 .  Was he mistaken, or
are we talking about completely unrelated things?

Best,

> Could you clarify this, please?  Do you mean that darcs check in 1.0.9
> was able to detect inconsistencies that previous darcs versions did  
> not?
> Or do you mean that 1.0.9 introduced some new corruption?

The former.  Immediately after installing darcs 1.0.9, the very next  
thing I did was run "darcs check" in several different repositories  
on two different computers.  They reported inconsistencies.  Arno (I  
think) experienced something similar.

>> Is there any known issue that would cause this?  I don't remember
>> exactly what version of darcs I last used on these repos.  Probably a
>> 1.0.9 pre-release.
>
>> Arno was using 0.9.19.  :-)
>
> I thought in issue517, he said he was using 1.0.9 .  Was he  
> mistaken, or
> are we talking about completely unrelated things?

The previous version of darcs that he was using before he installed  
1.0.9 was 0.9.19.

Regards,

Zooko

On Sat, Aug 11, 2007 at 12:31:47PM -0600, zooko wrote:
> >Could you clarify this, please?  Do you mean that darcs check in 1.0.9
> >was able to detect inconsistencies that previous darcs versions did not?
> >Or do you mean that 1.0.9 introduced some new corruption?
> 
> The former.  Immediately after installing darcs 1.0.9, the very next  
> thing I did was run "darcs check" in several different repositories  
> on two different computers.  They reported inconsistencies.  Arno (I  
> think) experienced something similar.

These would be very interesting repositories to look at!
-- 
David Roundy
Department of Physics
Oregon State University

These ones came from my parallels virtual machine.  They are suspect  
because all kinds of weirdness happened, such as segfaults,  
filesystem corruption of the NTFS filesystem.

http://zooko.com/zfec.tar.bz2
http://zooko.com/tahoe-broken-filesystem_0.tar.bz2
http://zooko.com/pyutil-from-parallels.tar.bz2

However, these ones came from my Toshiba real physical made-of-atoms  
laptop, which has displayed no such problems, and these ones have  
similar "inconsistent" results from "darcs check":

http://zooko.com/tahoe-p.tar.bz2
http://zooko.com/pyutil-from-zaulaw.tar.bz2
http://zooko.com/zfec-from-zaulaw.tar.bz2

Regards,

Zooko

On Tue, Aug 14, 2007 at 10:26:16 -0600, zooko wrote:
> http://zooko.com/tahoe-p.tar.bz2
> http://zooko.com/pyutil-from-zaulaw.tar.bz2

Argh, these appear to be due to issue348 (for which, sorry)

Repair trivially fixes them.  The check says
  Looks like we have a difference...
  Nothing

which isn't too helpful

Marking as duplicate.  Have submitted a patch to not create those darcs-ssh
directories in the wrong place.