Package: apt
Version: 0.7.9
Severity: wishlist
Tags: patch
Hi,
It seems that currently apt does not convert the Packages or Sources
file to the codeset from the terminal. This is probably because there
never was and still isn't a requirement that the debian/control
file is written in a certain encoding. However most packages
do have their control file written so that they're compatible with UTF-8.
Apt also currently already converts the long description from UTF-8
to the current codeset.
It seems there currently are 11 packages that aren't compatible with
UTF-8:
http://lintian.debian.org/reports/Tdebian-control-file-uses-obsolete-national-encoding.html
There is also a release goad that tries to get rid of those, see
bug #453954.
I've attached a patch that tries to convert it from utf-8 to the
currently selected codeset, and return the original string if it fails.
One effect this change has is that descriptions for any of those packages
now show up fully instead of being truncated at the first char that
isn't UTF-8.
I might not have written a patch that's does the changes at the right
place because I'm not familiar with the code. But the changes work for
me.
Kurt
Subject: Re: apt: Convert Packages/Source file from utf-8 to codeset.
Date: Sun, 2 Mar 2008 15:42:06 +0100
Hi,
Can someone please comment on this patch and maybe apply it?
We're now in a state where all control files are in UTF-8.
Kurt
On Sat, Dec 22, 2007 at 08:34:06PM +0100, Kurt Roeckx wrote:
> Package: apt
> Version: 0.7.9
> Severity: wishlist
> Tags: patch
>
> Hi,
>
> It seems that currently apt does not convert the Packages or Sources
> file to the codeset from the terminal. This is probably because there
> never was and still isn't a requirement that the debian/control
> file is written in a certain encoding. However most packages
> do have their control file written so that they're compatible with UTF-8.
> Apt also currently already converts the long description from UTF-8
> to the current codeset.
>
> It seems there currently are 11 packages that aren't compatible with
> UTF-8:
> http://lintian.debian.org/reports/Tdebian-control-file-uses-obsolete-national-encoding.html
>
> There is also a release goad that tries to get rid of those, see
> bug #453954.
>
> I've attached a patch that tries to convert it from utf-8 to the
> currently selected codeset, and return the original string if it fails.
>
> One effect this change has is that descriptions for any of those packages
> now show up fully instead of being truncated at the first char that
> isn't UTF-8.
>
> I might not have written a patch that's does the changes at the right
> place because I'm not familiar with the code. But the changes work for
> me.
>
>
> Kurt
>