Bug#701081: debian-policy: mandate an encoding for filenames in binary packages
tag 701081 patch
I think that it emerges from the discussion that there are good uses of
Unicode, and that somebody would need to step up and ensure that a dozen of
packages are corrected if we were to restrict further the encoding of file
names. Moreover, there seems to be a good self-discipline, and Unicode is
not used in paths that are central on non-Unicode systems.
Given that currently the Policy does not mention anything about file names, I
think that it would be fair to fill the gap by documenting the use of Unicode
as current practice and recommend ASCII for most cases. This does not preculde
further restrictions if needed. I volunteer to contact the maintainer of
lletters-media and ooohg, the only packages with non-Unicode file names.
I attached a slightly updated patch. I have not added that the policy is for
'the files that have been created after the binary package is "Installed"',
because I think that it is clear throughrough chapter 10 that "installed files"
means this. Nevertheless, it would be nice to have such a definition black on
white somewhere else, to be discussed in another thread.
Have a nice week-end,
>From e0d128e11506ca142999bc79104cd1442d0105df Mon Sep 17 00:00:00 2001
From: Charles Plessy <firstname.lastname@example.org>
Date: Sun, 24 Feb 2013 12:11:00 +0900
Subject: [PATCH] Installed file names must be in UTF-8 and should use only
policy.sgml | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/policy.sgml b/policy.sgml
index d81891c..c619d44 100644
@@ -9463,6 +9463,16 @@ done
+ <sec id="filenames">
+ <heading>File names</heading>
+ The name of the files and directories installed by binary packages
+ must be encoded in UTF-8 and should be restricted to ASCII when they
+ can be represented in that character set.