genuine optional (used by one/some protocol) or things common with all protocols, and
needed by users of all protocols. Last ones *must* be handled by abstraction (we must
maintain all protocols, and without coding everything 12 times), in former case abstraction
is useful.
Message identification is obviously necessary for all protocols, for delivery reports. This
is way I called for abstraction here. This has nothing to do with conservatism, I want only
avoid duplicate (many times over) work
An example how to do this:
a) only smsc knows how message identification is generated. Outside, id is just a given
string.
b) how smsc constructs depends on configuration. Configuration variable message-
identification lists fields used:
For instance, with SMPP you can write:
message-identification = msgid
But with other protocols you cant do better than:
message-identification = timestamp, from-number, to-number
c) If you do not want delivery reports at all, say
message-identifcation = none
I know that somebody will say "my boss want support for SMPP, he do not care a s*t of
others". But consider this: with abstraction a bugfix in one protocol will fix other proto-
cols, without it all protocols will be fixed separately. Your boss will understand this, if
he has ever coded more than few lines.
So, J�rgen, can start implementation of all protocol features by separating optional and
common ones ?
Aarno
On 11.2.2005, at 01:12, Stipe Tolj wrote:
J�rgen Thomsen wrote:OK. Here is one practical issue, which deliberately has been turned down: implementation of complete official protocol standards.
ok. This is the "utopia" we are looking to gain. But how to achieve it in a practical abstraction and real world?
Now, I don't want to be picky. But it's alway easy to put critics about "present states". That's not enough. As it was never for real life problems and issues. Pragmatic thinkers of this world do not only state the lacking of present states, they also imply on how (at least may be) a better solution is to be drawn.
So, how about giving a concret example on how you would do certain issues?
Stipe
mailto:stolj_{at}_wapme.de ------------------------------------------------------------------- Wapme Systems AG
Vogelsanger Weg 80 40470 D�sseldorf, NRW, Germany
phone: +49.211.74845.0 fax: +49.211.74845.299
mailto:info_{at}_wapme-systems.de http://www.wapme-systems.de/ -------------------------------------------------------------------
