[asterisk-dev] Zap channel naming is way too confusing

[Steven Critchfield]

On Thu, 2006-02-23 at 10:24 -0600, Tilghman Lesher wrote:
> I've thought about this a little bit more, and I think we need to have a
> channel name format which is significantly different, in order to make
> it absolutely clear that it is specified as span/relative-channel,
> rather than absolute-channel.  Just as a possible example, we should
> continue to have absolute-channel specified as Zap/nnn, whereas the
> span/relative-channel might be specified as Zap[ss/rrr].  This
> notation makes it much more clear as to the meaning of the channel
> specification.  In fact, just for clarity of the translation, we might
> always specify the absolute channel, in a format such as:
> Zap[ss/rrr](nnn), e.g. Zap[3/4](52) or Zap/52.

First opinion is yuck.  Not all Zap hardware has spans. Consider the
analog cards, are you going to number them as a span and a channel at
once? 

Think of future hardware that might decide to go the zap route for a
driver. Think of a future if we have a T3 card, where the current idea
of a span is just a small portion of the true span. Then are we going to
augment the format to Zap[TT/SS/NN] To handle T3 span, T1 span, channel
number?

While I admit that Zap/Chan is going to cause the same troubles when you
think out further to multiple T3 spans. 

Neither way helps when you contemplate the idea of hot swappable cards. 

Maybe we need to be able to expose up what card the channel is on and
then further a span and channel. 

No real answers here, but a few more thoughts on what needs to be
considered before we commit to writing any code.
-- 
Steven Critchfield <critch at basesys.com>