asterisk users - Jul 2009 - chan_dahdi.conf parser question

My config works fine but I must be missing a concept because a small change 
gives an unexpected result. Can someone help me understand the 
chan_dahdi.conf parser that would explain this?

Based on the config below, Channels 1-23 are assigned to the context 
inbound-pri, and Channels 25-47 are assigned to the context outbound-pri. As 
expected.

[trunkgroups]
; We do not do NFAS at this time

[channels]
echocancel = yes
switchtype = national

context = outbound-pri
signalling = pri_net
group = 2
channel => 25-47

context = inbound-pri
signalling = pri_cpe
group = 1
channel => 1-23


HOWEVER when I simply reverse the order of the channels on the last few 
lines like so:

[trunkgroups]
; We do not do NFAS at this time

[channels]
echocancel = yes
switchtype = national

;This A part (4 lines) was swapped with the B part
context = inbound-pri
signalling = pri_cpe
group = 1
channel => 1-23

;This B part (4 lines) was swapped with the A part
context = outbound-pri
signalling = pri_net
group = 2
channel => 25-47


UNEXPECTED:
Channels 1 is unexpectedly assigned to the context outbound-pri
Channels 2-23 are 'properly' assigned to the context inbound-pri
Channels 25-47 are 'properly' assigned to the context outbound-pri

This is not a dahdi reload problem.  Even a complete system reboot returns 
these results
Am I missing a parser concept that would make this rational or is this an 
actual defect?
1.6.0.10, & Dahdi 2.2.0

For what it's worth I notice that in the sample chan_dahdi.conf file 
signalling is usually spelled with two L's but at least once spelled with 
just one L.  Both are correct in canonical human English, but does the 
parser understand them both?

Sample chan_dahdi.conf Line 809:
;signaling=em
;channel => 15
;channel => 16

-Karl

On Tue, 2009-07-28 at 15:32 -0500, Karl Fife wrote:
> My config works fine but I must be missing a concept because a small change
> gives an unexpected result. Can someone help me understand the 
> chan_dahdi.conf parser that would explain this?
I'll do my best.
> Based on the config below, Channels 1-23 are assigned to the context 
> inbound-pri, and Channels 25-47 are assigned to the context outbound-pri.
As
> expected.
So far, so good.
> HOWEVER when I simply reverse the order of the channels on the last few 
> lines like so:
> 
> [trunkgroups]
> ; We do not do NFAS at this time
> 
> [channels]
> echocancel = yes
> switchtype = national
> 
> ;This A part (4 lines) was swapped with the B part
> context = inbound-pri
> signalling = pri_cpe
> group = 1
> channel => 1-23
> 
> ;This B part (4 lines) was swapped with the A part
> context = outbound-pri
> signalling = pri_net
> group = 2
> channel => 25-47
> 
> 
> UNEXPECTED:
> Channels 1 is unexpectedly assigned to the context outbound-pri
> Channels 2-23 are 'properly' assigned to the context inbound-pri
> Channels 25-47 are 'properly' assigned to the context outbound-pri
That *is* unexpected.  If this can be reproduced on your system, please
open an issue report at http://issues.asterisk.org/ as this is not the
intended behavior.
> For what it's worth I notice that in the sample chan_dahdi.conf file 
> signalling is usually spelled with two L's but at least once spelled
with
> just one L.  Both are correct in canonical human English, but does the 
> parser understand them both?
That was one of my pet peeves with Asterisk 1.4 and earlier.  I asked
the developers to address it, and it's my understanding that in 1.6.x
and later that Asterisk will accept the word "signaling" with either
one, two, or even three 'l's.  :-)

-- 
Jared Smith
Training Manager
Digium, Inc.