[issue1272] update bug when file mode changes

Adrian Buehlmann mercurial-bugs at selenic.com
Thu Aug 21 14:59:40 CDT 2008


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

Reported by Marko Käning in
http://selenic.com/pipermail/mercurial/2008-August/020931.html

Diagnosed by Matt Mackall

On 20.08.2008 20:04, Matt Mackall wrote:
> Here's a new test:
> 
> #!/bin/sh
> rm -rf a
> hg init a
> cd a
> echo 1 > a
> hg ci -Aqm1
> hg clone -q . f
> cd f
> echo 2 > a
> chmod +x a
> sleep 1
> hg st -A
> hg up --debug
> hg st -A
> cat a
> 
> And here's what it produces:
> 
> M a
> resolving manifests
>  overwrite False partial False
>  ancestor 8595fb912127 local 8595fb912127+ remote 8595fb912127
>  a: update permissions -> e
> 0 files updated, 0 files merged, 0 files removed, 0 files unresolved
> C a
> 2
> 
> When we update the permissions, we're marking the file clean.


On 20.08.2008 20:35, Matt Mackall wrote:
> On Wed, 2008-08-20 at 20:23 +0200, Marko Käning wrote:
>> That means there is a bug?!?
> 
> Yes.
> 
> To trigger it you need to:
> 
> - get a file in the unset state (clone or update) 
> - change its contents and its exec flag without checking status
> - do an update that changes just the file's exec flag 
> 
> The latter typically means doing a "backwards update", ie from modified
> tip back to tip.

----------
messages: 6826
nosy: abuehl, mpm
priority: bug
status: unread
title: update bug when file mode changes

____________________________________________________
Mercurial issue tracker <mercurial-bugs at selenic.com>
<http://www.selenic.com/mercurial/bts/issue1272>
____________________________________________________



More information about the Mercurial-devel mailing list