Re: charset question (Greek)

----- Original Message -----
From: "Richard Braakman" <[email protected]>
To: <[email protected]>
Sent: Wednesday, March 13, 2002 11:11 AM
Subject: Re: charset question (Greek)


> On Wed, Mar 13, 2002 at 10:22:40AM +0100, Andreas Fink wrote:
> > I would suggest that recoding doesnt occur at all unless its necessary.
> > This means that every SMSC driver would need to be aware of all
> > coding types and recode once from source to target if needed.
>
> Unfortunately you can't do sms-split right without knowing the target
> character set.  Probably the smsbox will have to convert to the specified
> data coding (such as GSM charset) before splitting.  This might mean that
> the SMSC driver then has to convert back to iso-latin-1 or something in
> order to satisfy the SMSC protocol.
>
> A related problem is that the smsbox might not always know what the
> maximum sms size is.  Historically it is predictable, but it seems that
> for example SMPP is in the process of dropping these size limits.
> This might mean that the real solution is to push sms splitting all
> the way to the SMSC drivers, and provide library functions to avoid
> duplicated code.  But that means that the SMSC drivers would have to
> know about smsbox configurations such as split-chars and headers and
> footers...  It might simplify delivery notification, though, if an
> sms isn't split until the last moment.  On the other hand it might
> mean passing huge messages through the system because max-messages
> isn't applied until the last step.

I vote and offer me to do the work to always use utf-8 inside kannel, do
the recoding in smsbox to accept other codings at http (app -> smsbox)
(like now, the text=..&charset=iso-8859-1 or defining the charset in http
post) and probably defining a default charset to smsbox->app, like my
mo-recode, to simplify applications (if possible, recode to iso8859-1, for
example).

Then, smsc_* code should know which charset to use and know how to
split the message.

But I guess it's better to do a release of kannel and start to think in a
new
archictecture, using the modularity that have been spoken.

There's some issues I've been thinking to new architecture, like having a
sms_router process, some smsc_* processes that connects to it, informing
its capabilities (like sms lenght and accepted charsets), and have smsbox
(http process) connecting to it and inform what services does it provide.

Or having a monolith process.

And have kill -HUP reload the configuration.

Well, maybe I'm just rambling because of the lack of sleep :)

>
> --
> Richard Braakman
> Will write free software for money.
> See http://www.xs4all.nl/~dark/resume.html
>
>