Re: stat: possible feature retraction: speak now or hold your peace

[Dmitry V. Levin]

On Fri, Nov 25, 2005 at 10:18:25AM +0100, Jim Meyering wrote:
> [FYI, I posted a similar message just before making the incompatible
>  change, but no one objected. ]
> 
> Due to complaints that the new (5.93) stat has broken some Debian scripts,
> I'm thinking of making the following changes in coreutils-5.94 and
> on the trunk:
> 
>     stat's --format=FMT option now works the way it did before 5.3.0:
>     FMT is automatically newline terminated.  The first stable release
>     containing this change was 5.92.
> 
>     stat accepts the new option --printf=FMT, where FMT is *not*
>     automatically newline terminated.
> 
>     stat: backslash escapes are interpreted in a format string specified
>     via --format=FMT or --printf=FMT.
> 
> I have mixed feelings about doing this.  On one hand, the change has
> been in test releases for a pretty long time.  On the other, it's been
> in stable releases for only a month.
> 
> If you know of scripts that rely on the current (5.3.0-5.93) behavior
> and feel strongly that we should retain the current behavior, speak up now.

Since >=5.3.0 is deployed for a long time yet (i.e. I packaged first
testing coreutils package based on 5.3.0 branch for ALT Linux Sisyphus
about 15 months ago), scripts that rely on the current behavior are
certainly present.  At least one script I found uses stat in the
following form:
sizes=($(find somepath -print0 |xargs -r0 stat --format='%s '))
I believe that --format behavior change will break this script.


-- 
ldv

pgpS_xuI54Cnw.pgp
Description: PGP signature