Re: Working copy text base is corrupt?

Patrik Husfloen <pathu440@student.liu.se> writes:

> not that it's very useful but here is the result :)
>
> 0000C560: 0D 04
> 0000C561: 0D 03
> 0000C562: 0D 08
> 0000C563: 0D 03
> 0000C564: 0D 05
> 0000C565: F9 03
> 0000C566: 12 03
> 0000C567: 07 03
> 0000C568: 00 03
> 0000C569: 00 03
> 0000C56A: 00 03
> 0000C56B: 00 03
> 0000C56C: 1E 03
> 0000C56D: EB 01
> 0000C56E: 03 00
> 0000C56F: 15 FC
> 0000C570: 12 FC
> 0000C571: 12 FC
> 0000C572: 12 FC
> 0000C573: 12 FC
> 0000C574: 04 FC
> 0000C575: 03 FC
> 0000C576: 08 FC
> 0000C577: 03 F9
> 0000C578: 0A 12
> 0000C579: 03 07
> 0000C57A: 03 00
> 0000C57B: 03 00
> 0000C57C: 03 00
> 0000C57D: 03 00
> 0000C57E: 03 1E
> 0000C57F: 03 EB

If I interpret the output format correctly, this implies that the
corruption is limited to 32 consecutive bytes. It doesn't look random
either, bits of the second column (C560 to C56B and C578 to C57F)
almost match bits of the first column (C574 to C57F and C566 to C56D).

You stated in an earlier mail that the filesystem time stamp on the
corrupt text base matched the time stamp in the .svn/entries file.
This seems to imply that the corruption occurred when the file was
installed. However the checksum is calculated at that time, so it's
hard to see how it could be wrong. Mysterious.

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org