Re: [gmime-devel] Encoding headers: Is UTF-8 a sane default now?

[Michael Gratton <mike vee net>]

On Thu, Jun 30, 2016 at 9:05 PM, Jeffrey Stedfast <fejj gnome org> 
wrote:
> On 6/29/2016 10:36 PM, Michael Gratton wrote:
> > Thanks for the feedback, guys.
> >
> > So in terms of init'ing GMime to do the right thing: There's nothing 
> > to do for US-ASCII headers that will always be used if valid to do 
> > so, as long as GMIME_ENABLE_USE_ONLY_USER_CHARSETS is not passed to 
> > gmime_init() it will do the same thing for iso-8859-1, and if I call 
> > g_mime_set_user_charsets() with just UTF-8, it will use that for all 
> > other header strings.
> >
> > If I do set the user charsets to UTF-8 then I assume that means all 
> > headers will be sent as one of those three - i.e. no 
> > non-ascii/iso-8859-1 header strings will ever be encoded with 
> > something like iso-8859-*, windows-1251 or Shift-JIS, right?
> >
> > //Mike
> I think you're correct, yes.

Cool, thanks.

> The gmime-charset-map-private.h list of charsets is used as a way of 
> figuring out which charsets the header values *could* be encoded into 
> and the user-charsets list is used to further filter that down to an 
> "allowed-to-use" subset.
>
> The user-charsets list is also used as a list of charsets to try to 
> convert from when it encounters undeclared 8bit text in the header 
> values.
>
> FWIW, I've removed iso-8859-7 from the list and regenerated the 
> tables in git master, so that should solve the iso-8859-7 issue that 
> your user has hit.

I did see that, ta. I've had to add a workaround anyway in Geary though 
since it'll be a while before it'll hits the distros.

> I'd be happy to remove more of them, I just want to be careful about 
> doing that until we get data for those regions that the iso charsets 
> have been deprecated and/or we know that the vast majority of clients 
> used by users in those regions support UTF-8.

Well if it's any indication, Geary seems to have a number of users from 
around the world and while it always sends message bodies as UTF-8 I 
haven't been able to find a bug complaining about that. In fact, this 
issue of sending headers not using UTF-8 is the only thing I have heard.

//Mike

--
⊨ Michael Gratton, Percept Wrangler.
⚙ <http://mjog.vee.net/>