[Mozilla] Fwd: Please consider if your locale could default to UTF-8 for outgoing email

Jordi Serratosa jordis.lists a gmail.com
div gen 31 14:37:16 CET 2014

A la llista de traducció de Mozilla comenten que cada llengua caldria 
considerar si vol canviar la codificació de caràcters per defecte a 
UTF-8 en l'enviament de missatges de correu.
Si voleu seguir la conversa, és aquí:

Actualment, el català té configurat que els missatges s'enviïn per 
defecte en ISO-8859-1.
En principi jo no veig inconvenients per fer el canvi, que sembla prou 
lògic, però no sé si algú en sap més i creu que no seria bona idea...


jordi s

-------- Missatge original --------
Assumpte: 	Please consider if your locale could default to UTF-8 for 
outgoing email
Data: 	Fri, 31 Jan 2014 12:30:13 +0200
De: 	Henri Sivonen <hsivonen a hsivonen.fi>
A: 	dev-l10n a lists.mozilla.org, dev-apps-thunderbird a lists.mozilla.org

Yesterday, the default character encoding for new outgoing email in the
en-US localization of Thunderbird changed from ISO-8859-1 (which really
means windows-1252 labeled as ISO-8859-1) to UTF-8. (Thanks Magnus!)
Previously, many other localizations had already set the default
character encoding for new outgoing email to UTF-8. This localizations are:
South Ndebele
Southern Sotho
Traditonal Chinese

Note that this isn't just a matter of African languages that have to use
UTF-8 using UTF-8. The above list includes languages that do have
Thunderbird-supported legacy encodings (Arabic, Belarusian, Greek,
Farsi, Macedonian, Polish, Romanian, Russian, Serbian, Vietnamese,
Traditional Chinese).

I encourage other localizations to consider whether they'd be ready to
change the default for outgoing email (mailnews.send_default_charset in
messenger.properties) to UTF-8.

The only reason not to would be regionally still-popular UTF-8-incapable
email clients. (Unlikely at this day and age.)

In particular, since the Belarusian, Macedonian, Russian and Serbian
localizations have already been able to make the switch to UTF-8, it is
very probable that the Bulgarian and Ukrainian localizations could do
so, too, at this time.

I'd also like to especially encourage the German, Italian and Breton
localizations to consider if they could make the switch to UTF-8. These
three locales currently default to ISO-8859-15 for outgoing email. This
doesn't make sense, because:

  1) ISO-8859-15 is a younger encoding than UTF-8. If you are worrying
about UTF-8 being too new to be supported, on newness grounds, you
should worry even more about ISO-8859-15!

  2) Compared to ISO-8859-1, ISO-8859-15 was motivated by three things:
less common French letters mysteriously omitted from ISO-8859-1,
transliterations in the Finnish context (totally missing the point of
transliterations!) and the euro sign. When the French and Finnish
localizations as well as many other localizations used in the Eurozone
don't need to default to ISO-8859-15, German, Italian and Breton almost
certainly don't need to, either. But instead of changing back to
ISO-8859-1, let's treat this as an opportunity to switch to UTF-8.

- -

Currently, Thunderbird already silently upgrades a message to UTF-8 even
when the default is something else if the message contains characters
that can't be expressed in the preferred legacy encoding, so it's not
like the defaults prevent people from writing stuff in email. So why
bother changing the default then?

The reason why Thunderbird doesn't always use UTF-8 for outgoing email
is that way back when UTF-8 support was initially added to email
clients, older email clients out there didn't all already supports
UTF-8, so it makes sense to use already-supported legacy encodings when
possible. But that was a long ago. UTF-8 was invented over 20 years ago
and at least for the past decade, the ability to receive UTF-8 has been
commonplace. We should stop pretending that UTF-8 avoidance still adds
value when pretty much everyone already uses email clients capable of
receiving UTF-8. The Gaia email client already simply uses UTF-8 for all
outgoing email.

The UTF-8 avoidance capabilities don't come for free. In addition to
having to maintain encoders for legacy encodings (which needs to happen
to support form submissions in Firefox anyway) and having code
complexity to deal with the silent upgrade to UTF-8 (which isn't that
much additional code complexity considering what Firefox form
submissions require anyway) it means having to maintain the capability
of the HTML serializers (for HTML mail) to output encodings other than
UTF-8 and it comes with the UI burden (both in terms of adding more
stuff that the user may end up thinking about and in terms of adding a
ball and chain legacy to code maintenance) since we seem to feel that if
there is the possibility to use encodings other than UTF-8, there should
be UI for it. (My primary interest in this is the inertia that
Thunderbird having UI for this imposes on making encoding-related
changes to Firefox.)

So the two previous paragraphs seem to argue for removing the
configurability and just always sending UTF-8. What's this about
localizations changing the default configuration then?

Changing the default configuration is a first step that makes reducing
configurability and the associated code complexity less scary in the
future. Making a change like this product-wide regardless of locale
tends to bring up concerns that maybe the email client environment for
some locale out there isn't ready yet. As more and more locales make the
call that they can make the switch and default outgoing email to UTF-8,
the concern of the form "Where *I* live, we don't have problems with
this but I *worry* that *they* over *there* do." becomes less of a
source of inertia.

Henri Sivonen
hsivonen a hsivonen.fi
dev-l10n mailing list
dev-l10n a lists.mozilla.org

-------------- part següent --------------
Un document HTML ha estat eliminat...
URL: <http://llistes.softcatala.org/pipermail/mozilla/attachments/20140131/65711c9b/attachment.html>

Més informació sobre la llista de correu Mozilla