[Buildd-tools-devel] Bug#500871: Bug#500871: schroot: Revisiting bug#427047: followup implementation question

[Roger Leigh]

On Thu, Oct 02, 2008 at 10:24:13AM +0200, Peter Rabbitson wrote:
> I finally came back to using schroot, and was delighted to find the
> feature I proposed some time ago fully implemented.

You're welcome!  I know it took some time to finally implement this; it
required some thought and finding enough free time to do, including
writing schroot-mount to do it.

> I have a
> documentation/example question though. Everywhere in the documentation
> and in the default fstab file you assert that "The format of this file
> is the same as  for  /etc/fstab,  documented in fstab(5)". However in my
> initial proposal I dropped the 5th and 6th field, namely fs_freq and
> fs_passno. Is there a reason to maintain those without mentioning that
> their values are ignored? Or are the values in fact interpreted in some way?

We use the standard POSIX getmntent family of functions to read the
file; these are the same functions used to read /etc/fstab and
/etc/mtab.  As a result, the file format is identical, and this requires
using all 6 fields.  This is more reliable than parsing the file
ourselves--we use the system parser that all other programs use.

We don't mention that the values are ignored; this is partly because you
have to write values as you would in /etc/fstab (i.e. 0 0), and partly
because this gives us the ability to use them in the future if we wanted
to run fsck on session startup, for example.

Regards,
Roger

-- 
  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux             http://people.debian.org/~rleigh/
 `. `'   Printing on GNU/Linux?       http://gutenprint.sourceforge.net/
   `-    GPG Public Key: 0x25BFB848   Please GPG sign your mail.