Re: [Gnu-arch-users] BUG: feature request: 'tla chmod' which 'touch'es files

[John Meinel]

Aaron Bentley wrote:
> Zenaan Harkness wrote:
>
> > Too many times I've been bitten by chmod not 'touch'ing the
> > corresponding file, and therefore not propagating the perms change into
> > the commit.
> >
> > A function like the following would probably do the trick:
>
>
> > Anyway, one to add to the feature list.
>
> I'd rather fix revlibs so permission changes are detected, instead.
>
> Aaron

Well, fixing revlibs to detect the change is certainly important. But I 
was thinking that if you accidentally used an editor that didn't break 
hardlinks, then you would at least have a readonly file so you would 
realize you shouldn't edit it, or at least should break the hardlink first.

Is there a specific reason why permission changes are not detected? Is 
it just that timestamps are checked first, and if they are the same, no 
further work is done?

So is it just the matter of checking permissions as well as checking 
timestamp?

Also, if checking timestamp, which one are you checking? As I recall, 
mtime is the last modification to the file, but ctime is the last update 
of permissions. (Although I think on Win32 ctime is the creation time)

John
=:->

signature.asc
Description: OpenPGP digital signature