Re: [Emacs-diffs] master 83c3a61: lisp/gnus/message.el (message-send-mail): Don't wrongly assert on Unicode msg

Re: [Emacs-diffs] master 83c3a61: lisp/gnus/message.el (message-send-mail): Don't wrongly assert on Unicode msg

[Stefan Monnier]

>     the message seding is handled by an external program, so no Unicode
>     encoding is performed in Emacs. Thus Emacs must not assert that the
>     Unicode encoding was performed.

Not sure what this has to do with Unicode.  It seems to just be about
encoding or not encoding: whether encoding is done with one of the
Unicode coding-systems (e.g. utf-8) or something else (e.g. latin-1)
doesn't seem to make a difference.

Or am I missing something?


        Stefan

Re: [Emacs-diffs] master 83c3a61: lisp/gnus/message.el (message-send-mail): Don't wrongly assert on Unicode msg

[Oleh Krehel]

> >     the message seding is handled by an external program, so no Unicode
> >     encoding is performed in Emacs. Thus Emacs must not assert that the
> >     Unicode encoding was performed.
>
> Not sure what this has to do with Unicode.  It seems to just be about
> encoding or not encoding: whether encoding is done with one of the
> Unicode coding-systems (e.g. utf-8) or something else (e.g. latin-1)
> doesn't seem to make a difference.
>
> Or am I missing something?

I'm sending an SMIME-encoded message using openssl. This requires to
keep the plain text non-encoded.

There's already existing message.el code that sets
`message-inhibit-body-encoding' to t. But if the encoding is
inhibited, we should not assert if Unicode was not encoded. Thanks to
this patch I can now send SMIME-encoded email that has e.g. German
umlauts in it.

I found the existing support for sending SMIME-encoded email lacking.
For a while, some mail clients would detect the sent email as signed,
but others not. I can contribute the openssl signing code that works
for me now if there's interest.

regards,
Oleh

Re: [Emacs-diffs] master 83c3a61: lisp/gnus/message.el (message-send-mail): Don't wrongly assert on Unicode msg

[Stefan Monnier]

> There's already existing message.el code that sets
> `message-inhibit-body-encoding' to t.  But if the encoding is
> inhibited, we should not assert if Unicode was not encoded.

So from your description I think indeed this has nothing to do with
Unicode: the code you changed doesn't "assert if Unicode was not
encoded".  It asserts that the text was not encoded.  Unicode or
anything else.  German umlauts are chars that indeed appear in the
Unicode charset, but they also appear in many other charsets (such as
latin1).

"Encoding" turns chars into bytes and the test verifies that there are
only bytes left.  


        Stefan