Re: [Quilt-dev] Quilt 0.65 coming soon

[Andreas Grünbacher]

2016-10-24 15:45 GMT+02:00 Jean Delvare <address@hidden>:
> On Mon, 24 Oct 2016 18:56:20 +0900, Yasushi SHOJI wrote:
>> On Mon, Oct 24, 2016 at 4:30 PM, Jean Delvare <address@hidden> wrote:
>> > I added this short help text after noticing that the option was not
>> > documented at all. But the truth is, I don't use the quilt mail
>> > command, and I don't really know what the option does. It was an
>> > uneducated guess. You certainly know better. Would you mind providing a
>> > patch to describe the option properly? I'd be happy to apply it before
>> > releasing v0.65.
>>
>> Unfortunately, I don't use it neither. My comment was from 15 min. reading
>> and grepping of quilt source code and document with "charset" right before
>> writing translation.
>>
>> It seems the command and the option was added by the commit 0ae231e
>> by Andreas Gruenbacher <address@hidden> in 2005!
>
> That's what I found too, and the option has never been documented.
>
>> If we don't get from anyone using the command / option in a short while,
>> i'll try to write up a few lines.  Hoping someone beat me to it.
>
> My understanding is that --charset should match what is used in the
> patches themselves. The value is used for the Content-Type header of
> each message and there is no conversion done.

Indeed. On systems using UTF-8, that header should automatically be
set appropriately; on other systems, the ISO-8859-15 character set is
assumed.

Headers are supposed to be 7-bit only, so we need to encode recipient
addresses with special characters. (This still doesn't catch special
characters in the subject, but that doesn't seem to be a very bit
problem at least.)

> It is also passed to the edmail helper script, which uses it to encode
> the recipient names. But that is an internal detail which I don't think
> the user needs to know. In fact I don't know why edmail needs to know
> the charset used for the body, if the recipient fields end up being
> encoded anyway it might as well always use UTF-8? But that's a somewhat
> different topic, and I'm not familiar with email encoding so I could be
> plain wrong.

The Content-Type header still needs to be added to messages.

> Andreas, what would be a good description for the mail --charset
> option? "Specify the character encoding of the patches."?

The --charset option allows to specify a particular message encoding
on systems which don't use UTF-8 or ISO-8859-15.

Thanks,
Andreas