[issue1479] verify fails on copy data magic bytes

[Adrian Buehlmann]

New submission from Adrian Buehlmann <adrian at cadifra.com>:

Per the thread http://selenic.com/pipermail/mercurial/2009-January/023553.html

On 21.01.2009 18:09, Matt Mackall wrote:
> On Wed, 2009-01-21 at 11:34 +0100, Quenouille Guillaume wrote:
>> I tried to clone the repository, but this doesn't resolve the problem.
>> You will find attached to this email the file which raises problem. If
>> you commit it in Mercurial and then do a "hg verify", you will have
>> this integrity error. Is there a way to fix this ?
> 
> This is actually an obscure bug in verify. Your file starts with the
> magic bytes 01 0a which Mercurial uses internally to signal a revision
> contains copy data, so we escape them. So your file actually gets
> -stored as-
> 
> 01 0a 01 0a <actual file data>
> 
> ie an extra 4 bytes. In normal use, Mercurial properly strips off this
> prefix, but verify does things a little differently. We'll fix this in
> the next release.
> 
>> For the warning, a clone didn't resolved the problem.
> 
> I'm sorry, I meant a clone --pull. A local clone will simply hardlink
> all files, including the orphans.
>

----------
assignedto: mpm
messages: 8467
nosy: abuehl, mpm
priority: bug
status: unread
title: verify fails on copy data magic bytes

____________________________________________________
Mercurial issue tracker <mercurial-bugs at selenic.com>
<http://www.selenic.com/mercurial/bts/issue1479>
____________________________________________________