[horde] Greek in Applications other than IMP

[Otto Stolz]

Καλιμέρα,

Antonis A Plevrakis schrieb:
> How
> can in enter the following tag in every rendered horde page easily:
>    <meta http-equiv="Content-Type"
>    content="text/html; charset=windows-1253">

I do not deem that a good idea, because the WWW pages generated
by Horde contain much more than just the input fields for your
(presumably) Greek text. E. g., a list of e-mail subjects, or
a list of proper names (in an addressbook, say) may contain
characters from any conceivable script. The only charset that
can handle this is Unicode (in its various UTFs).


Furthermore, there is a pitfall with text input from WWW forms
(as used by Horde and its application modules): If the user enters
a character (via the keyboard, or by cut-and-paste) that is
not contained in the WWW form’s charset, the browser cannot
transmit it to the WWW server. In this case, the browsers
I have tested rather transmit the characters NCR¹, an HTML construct
which has no meaning whatsoever in other contexts (e. g. in e-mail,
or in a contact list).

Example: The user enters “ά”, an Alfa with Oxia, (rather than the
usual “ά”, an Alpha with Tonos), in a windows-1253 encoded WWW form.
The browser will send the string “#8049;” (six characters rather than
one!), or a similar beast, to the server; if this ends in an e-mail
message, the adressee will see “#8049;” rather than an Alfa, in the
middle of a Greek word. Imagine this sort of happening with every
vowel that carries an accent, a breathing-mark, or a Iota subscriptum:
it will render the text virtually incomprehensible.

The only feasible way to avoid this pitfall is to use Unicode-encoded
forms, throughout.


Hence, you’d be much better off if you’d configure your SQL
database so it would store all text fields in UTF-8.

We have done so when we upgraded to Horde 3. Now, any user
can enter his name, or his vacation text, or what else you can
think of, in his mother tongue, without any problems. And Horde
+ Imp take care of tagging the data according to the encoding
used, so addressees can correctly display them with any and all
e-mail clients (exept Eudora, as it sovereignly ignores vital
e-mail standards).

Good luck,
   Otto Stolz


_____________________

¹ NCR: cf. <http://www.w3.org/TR/html401/charset.html#h-5.3.1>.