I'm +1 for Rene-Alex. If bearerbox-host is part of smsbox group there is no reason for exclude the port. And i'm still thinking that 1-to-N architecture is no necessary broken taking this way.
M On 6/14/06, Rene Kluwen <[EMAIL PROTECTED]> wrote:
Okay... here's my 2 cts... I disagree with you, Stipe. smsbox needs to connect to the/a bearerbox. So the bearerbox-port should be a configuration parameter of smsbox. I will give you some logic: 1. smsbox connects to the tuple (bearerbox-host:bearerbox-port) 2. bearerbox-host is a configuration parameter of smsbox 3. ergo => bearerbox-port should be as well The core group is designed to serve bearerbox. A way-out-of-line argument for this is the use of smsbox for something completely different than connecting to the bearerbox. I leave possible uses to the creativity of the reader. smsbox -in this light- is a stand alone application. Not related to bearerbox at all. Conclusion: Both in the perspective of the user as well as in an architectural view, group = smsbox might contain a bearerbox-port value. Okay... so far the logic... I am self-respecting enough to come up with a possible solution. How about we inherit bearerbox-port from group = core BUT can be overridden by the same entry in group = smsbox. So the core section is only used if the same parameter in smsbox is omitted. This way, we also stay downward compatible. And everybody will be happy! Comments? Maybe I can raving too far here. But we can even make this a "general rule". Taking mutual parameters in group = core when not found in group = smsbox, wapbox, etc. I actually do this in other software. Rene Kluwen Chimit -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Stipe Tolj Sent: donderdag 15 juni 2006 2:33 To: Mi Reflejo Cc: devel@kannel.org Subject: Re: [PATCH] (sqlbox) sqlbox.patch Mi Reflejo wrote: > Good one, Alex > Adding bearerbox-port to smsbox group should be a good solution. > I'll follow that way. nop, I'm -0 on this... see my response on the other mail for argues. Alex argues that smsbox would be "runnable" without core group... now, that doesn't make sense in the current architecture. smsbox can't be running without bearerbox. It's a "I hook into an upstream box"-box and hence needs to have a connection to bearerbox. And that's why a core group is always semantically necessary. _If_ the smsbox could run without bearerbox, meaning you would launch it, and on the HTTP application layer side it could do things without having at the operational time a link to bearerbox, then the approach would make sense. But that's not the case. I recall again the 1-to-N architecture we have here for bearerbox(1)<->smsbox(N) connections. > Stipe, I'm waiting for your "sceleton build environment". yep, preparing.... we're in WorldCup feaver... so things need more time ;) Stipe ------------------------------------------------------------------- Kölner Landstrasse 419 40589 Düsseldorf, NRW, Germany tolj.org system architecture Kannel Software Foundation (KSF) http://www.tolj.org/ http://www.kannel.org/ mailto:st_{at}_tolj.org mailto:stolj_{at}_kannel.org -------------------------------------------------------------------
