Hello to all.

After several warnings but no one changing the code, I'm picking up the
job and finally fixing my own bug.

Very soon (I have already finished the code but I'm still testing and
documenting) I'll commit a patch to fix inconsistences in cgi
parameters (which were primarly caused by me).

When I commit, and in release 1.3.2 and next versions, we'll have:

MClass: 0 to 3 instead of 1-4
Coding: 0 to 2 instead of 1-3
MWI: 0 to 7 instead of 1-8

and internally, default values for every sms integer parameter will 
be -1, enabling:

MClass=-1 - undefined (if you don't give the parameter or set it to -1)
MClass=0-3 - real ETSI values

PID=-1 - don't put any value in SMSC PDU unless PID is required (like
in AT?)
PID=0-255 - enables you to say "I want a PID with a 0 value in PDU"

RPI: -1: undefined
RPI: 0 Reply Path disabled (in EMI is similar to -1, no value in pdu)
RPI: 1 Reply Path enabled (in EMI, "1" for "request")
RPI: 2 Reply Path enabled, equal to RPI=1 (in EMI, "2" for "reply");

Validity and Deferred: 
-1: don't set in pdu
0: doesn't make sense in practice, but it's allowed


dlr-mask = -1 or undefined - let kannel do what it wants 
dlr-mask = 0 - don't ask for dlr
dlr-mask = 1-31 - same as before


In general, the idea is that if you don't set a parameter, kannel can do
whatever it wants to do, choosing the best solution (for example,
choosing 7/8bits or UCS2 automatically), but if you say an explicit "0"
you are asking it to respect your choice. (obviously it won't try to
convert your 8bit data to 7bit if you say coding=7bit, but it will set
dcs to 7 bit anyway even if it gaves an error)




I've deprecated dlrmask and dlrurl query-string fields (they still work 
but you get Warnings on logs, just like flash parameter wish might
dissapear in next release). They are replaced to dlr-mask and dlr-url
to be consistent to other parameters (like X-Kannel-DLR-Mask in POST).




I'm also enabling Auth-Code (AC) input parameter (appl->smsbox) and 
HPLMN output parameter (smsbox -> appl), and only for EMI. The code 
for input and output is in smsbox.c and if other smsc's have some kind 
of Auth-Code output or HPLMN input, you'll just need to uncomment it.



I'd like to ear any comment to this but I guess it won't make any
discussion because this was already talked several times in the past.

Now it's the right time to fix what's wrong, just before releasing a new
version ;)



Reply via email to