>* bearerbox MUST be able to start with some (up to all) smsc connections
>failing and startup and should constantly retry to connect. that should
>probably resolved not on smsc level, but higher, to be global and
>systematic. emi2 and smpp seems to be ok, but cimd2 fails to start if cannot
>connect to smsc. why, you might ask? well .. I have a setup now where I have
>currently 11 smsc connections over internet and You can imagine that some
>network problems are often to occur, but a failing smsc must not prevent
>whole system from starting.

Totally agreed, and this has been discussed before, but I don't know
what decisions resulted.  Anyone?


>* as I have connections to smsc's in different countries, the number
>normalization does not work. I propose to rewrite unified-prefix and number
>normalization so that it would be based on smsc, assuming that a message in
>smsc2_rout() have it's smsc_id field filled. maybe a new config group -
>country - should be introduced, which will hold the number prefix
>normalization and all smsc's have a config param 'country' pointing to the
>country group.

Very good idea.  Again I'm surprised it hasn't been done already.


>* adding and reconfiguring smsc connections should be possible without
>restarting whole system.

+1 on that!  At the very least, a HUP signal should force a re-read of
the config file.




>* another idea, not very clear, however .. I have several connections to
>every smsc - because there are different numbers available for clients - and
>I have several operators - for example in estonia 3 operators and 3
>numbers - maybe a new parameter for smsc needs to be introduced - let's call
>it service-group for now, which will bind together (ie have same value) for
>all smsc connetions with similar service number. and then it should be
>easily possible to add option to enable routing only within the service
>groups.
>* it could be useful to add aliases (or constants - whatever we call them)
>into config files. 

This one might be controversial because it adds complexity to the program 
for unclear benefits, but it seems elegant in some ways too.

Steve

Reply via email to