[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#701081: debian-policy: mandate an encoding for filenames in binary packages


On Sun, 2013-02-24 at 11:54:01 +0900, Charles Plessy wrote:
> This could be done by an addition like the following, after section 10.9
> (Permissions and owners).  The wording is still a bit clumsy also, I am not
> sure if "installed" includes files created by maintainer scripts (which would
> be the intent here).  I named the section "File names", and not "File name
> character set", in case we would add other restrictions (such as length) in the
> future.

To make the installed situation pretty clear, it might make sense to
say something along the lines: «the files that have been created after
the binary package is "Installed"».

> +      <sec id="filenames">
> +       <heading>File names</heading>
> +
> +       <p>
> +         The name of the files installed by binary packages must be encoded in
> +         UTF-8 and should be restricted to ASCII unless there is a justified
> +         need for using other characters.
> +       </p>
> +      </sec>
> Some packages do not comply with the above.  Given the pace of the releases
> of the Policy, I am not sure that it is worth having first a should and then
> a must, if you or somebody else would have the time to tackle the issue
> after the Wheezy release.

I'd second something like this, but I'd first like us to consider if
we really want any non-ASCII characters in filenames. Currently on sid
there does not appear to be many such filenames (64 from my check, if
that's not bogus):

  $ LC_ALL=C zgrep '[^[:print:]]' \
    ftp.debian.org_debian_dists_sid_*_Contents-amd64.gz | wc -l

> By the way, how about directories ?

This is a matter of terminology, directories are also filenames, and
part of pathnames, which point to a directory instead of a file. I
don't see why we'd want to exclude directories from filenames.


Reply to: