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



On Wed 2016-06-29 09:59:53 -0400, Michael Gratton wrote:

Looking back through GMime bugs in b.g.o, it seems that in 2007 anyway 
using UTF-8 as a default encoding for headers was a bad idea. Is this 
still the case? I ask because a user has reported a Geary bug[0] that 
suggest for Greek at least ISO8859 is deprecated these days in favour 
of UTF-8, and indeed one mailer (Rainloop) seems to have problems with 
the older encoding.

fwiw, i agree that the default character encoding should be UTF-8, not
anything in the iso-8859-* family.

note that the raw headers themselves should probably still be all-ASCII,
though.  We're just talking about the example:

 =?iso8859-7?b?7OXm3fIg7OX03CDs8P3x4fI=?=

which would probably be better as:

 =?utf-8?b?zrzOtc62zq3PgiDOvM61z4TOrCDOvM+Az43Pgc6xz4I=?=

(both of these should yield the string "μεζές μετά μπύρας")

I can have Geary init GMime to default to UTF-8 for encoding non-ASCII 
headers, but is that a good idea today?

The only advantage of the 8859-* family over UTF-8 is shorter length.
For header fields (which are quite short already) i don't think this win
should trump the universality of UTF-8.

       --dkg


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]