Re: [ietf-types] Status of application/patch or text/patch?

On Jul 21, 2012 8:53 AM, Martin J. Dürst <duerst@it.aoyama.ac.jp> wrote:
>
> On 2012/07/21 5:49, Jon Moore wrote:
>>
>> On Fri, Jul 20, 2012 at 2:46 PM, Paul Libbrecht<paul@hoplahup.net>
 wrote:
>>>
>>> Erm... source charset makes sense to me (this is "the charset of the
patch").
>>> But target charset makes no sense into a media-type... this is "after
applying a patch" which is beyond the scope of a single file.
>>
>>
>> No, the issue is that the patch file could have mixed charsets. If the
>> change is 'convert README from iso-8859-1 to utf-8', then the patch
>> file itself will have two character sets:
>>
>> 1. The lines that begin with '-' or ' ' will be iso-8859-1.
>> 2. The lines that begin with '+' will be utf-8.
>>
>> Since these are incompatible charsets (neither is a superset of the
>> other), we need two charset parameters to describe them, and this
>> patch is not capable of being a text/* document. I'm open to other
>> names besides 'source' and 'target', but I'm not sure how we get
>> around having two of them for this case.
>
>
> Just use application/patch. I don't think current diff programs are able
to deal with character encoding parameters. I don't think people
necessarily need to know the source/target encoding in the above example,
unless something weird happens and they need to debug (in which case,
they'll find out by themselves through trial and error anyway).
>
> I also don't think there is much of a use case for:
> - Originator: Has files in Latin-1, makes a diff in that encoding,
>   sends to receiver.
> - Receiver: Has files in UTF-8, wants to apply patch and wants a
>   conversion from Latin-1 to UTF-8 to happen automatically.

I'm not saying that any conventional tools do this (though it would be nice
for them to grow into this functionality), but as someone who manages
public input on a fair number of docs, I receive a lot of 1252 or 8859
undermining my precious UTF-8 docs. If these are properly labeled, I can
iconv them our pull them into emacs and transcode them before applying
them.i expect that the burgeoning popularity of distributed version control
will accentuate this need as people become more adept at managing diffs.

> So I very much agree with Björn here.
>
> Regards,   Martin.
>
> _______________________________________________
> ietf-types mailing list
> ietf-types@ietf.org
> https://www.ietf.org/mailman/listinfo/ietf-types