[Asterisk-Dev] IAX spec: Text formats and character sets

[Tilghman Lesher]

On Saturday 30 April 2005 05:04, Olle E. Johansson wrote:
> > Since it's obvious that this is an unresolved issue, we should
> > avoid the issue at this juncture in the IAX2 spec and simply
> > specify that ASCII is the character format.  If at some point in
> > the future these arguments are resolved, then at that time a
> > revision may be made to the IAX2 specification allowing UTF-8 or
> > another character set.
>
> If you actually read my proposal, your proposal is in line with
> mine, but I limit it a bit more. I am saying that for
> extensions/dial strings, let's stick to numbers.

I'm afraid that's a non-starter; we already have support in IAX2 for
letters/numbers; rolling back the current featureset seems like a
sure way to cause all sorts of future incompatibilities.

> And for Caller ID 
> names outside of SIP, let's stick to whatever the PSTN standardize
> on. On top op this, add a layer that properly supports SIP UTF8
> international caller ID names and named extensions.

Hang on; we're talking about IAX2, not SIP.  Don't confuse the issue
by intermingling the two.  Just because SIP supports something doesn't
mean IAX2 has to.

> The additional proposal for IAX2 was to add a way to send full UTF8
> Caller ID names and extensions to be able to properly bridge to SIP
> and also enable full Caller ID names for those that have names
> outside of a-z.

That seems fine, as long as these are sent end-to-end and impose no
significant load on call processing.  However, I'm a bit concerned
about the code to translate this into a character set that is
functional on the PSTN.  Worse, do we then have to run this
translation on ALL CallerIDName strings sent to the PSTN, to avoid
something slipping through?

> We will have to add IDN (International Domain Names) support and
> character set conversion with ICONV regardless, if we want to be
> SIP and DNS compliant. Let's do it right, or simply remove chan_sip
> and forget all about it - then you're free to standardize on ASCII
> :-)

Again, this is IAX2.  We are NOT talking about SIP.

> I've been working with network communication software for over 15
> years, so I know that this is not simple to grasp or understand, I
> have full respect for that. I might have grown a bit bitter from
> all of these discussions that end with "I don't understand this, so
> let's stick with ASCII" leading to software we simply have to do
> "rm -r" on. Even Microsofts first XML notepad did not support
> anything else than a-z properly, even though XML was standardized
> in regards to character sets from the beginning ;-)
>
> My apologies if I offended anyone with my attitude, but I can
> simply can not accept "let's stick with ASCII" comments - it's very
> ignorant to a large part of the world, and  a large part of the
> Asterisk community.

To be clear, I'm not suggesting never implementing extended character
support; I just think that if we have multiple parties who are in
clear disagreement about what form this support should take, it is not
in our best interest to make a quick decision about it.  Otherwise, we
stand to cause fragmentation of the community, which is far worse than
delaying this important decision to some future date.

Perhaps some way can be found to allow each proposal to stand as an
experimental extension, and use several IEs, marked as experimental
extensions, such that experimental systems can continue to
interoperate with standard systems.  With this, we can permit each
proposal to be fully fleshed out in terms of implementation, and make
a decision that affects the standard definition of IAX2  (IAX2.1?) in
the future, based upon the merits of each system implementing distinct
character sets.  I'd propose using IEs 240-255 for this purpose.

-- 
Tilghman