[BXXPwg] Message Number Implementation issue

Gabe Wachob gwachob@wachob.com
Sat, 4 Nov 2000 01:48:00 -0500


Hi to you!

Where does the spec say that the msgno's need be unique only within a given
channel? I was looking all over for this, because, as you say, I thought I
heard some discussion about this (apparently, I remembered the delta
backwards). 

Section 2.2.1.1 (of the -07 framework I-D) says "the message number must
... have a different value than all other "MSG" messages for which a reply
has not been completley received". By glaring lack of any mention about
uniqueness within a Channel, I assumed that this meant that uniqueness
for message numbers was to be a requirement w/r/t to entire Session, not
w/r/t to each individual channel. 

So, I was confused because I thought the sentiment in this list was to change 
that, but I didn't see it reflected in the spec (or at least I couldn't find 
it).

	-Gabe


On Fri, Nov 03, 2000 at 10:34:11PM -0800, Marshall Rose wrote:
> hi. in the current beep spec, message numbers are unique per direction on a
> given channel. in the old bxxp spec, serial numbers were unique per
> direction on a given session.
> 
> one of the reasons for this change was the concern you cite regarding having
> to keep a global list of sequence numbers. the unconvincing response was
> that implementors could use a scheme along the lines of the one you mention.
> (the one i had in mind involved some bit shifting, but no division).
> however, the issue is now moot!
> 
> /mtr
> 
> ----- Original Message -----
> From: "Gabe Wachob" <gwachob@wachob.com>
> To: <bxxpwg@invisible.net>
> Sent: Friday, November 03, 2000 20:45
> Subject: [BXXPwg] Message Number Implementation issue
> 
> 
> > Hi-
> > Trudging through implementing PyBXXP, I've come across a little
> > optimization/simplification for generating message numbers that I'd like
> to
> > run by the folks.
> >
> > First of all, am I correct in the assumption that message numbers
> > must be unique across all channels now? I'm assuming the answer is yes (I
> > think this was discussed earlier). So, the question I have is whether
> > anyone sees a problems with assiging a subset of the 2147483647 possible
> > message numbers to each of the mandatory 257 channels? That is, can I
> > simply make message number spaces of size 2147483647/257 for each channel
> I
> > am working in? That way, I don't have to keep a global list of unresponded
> > msg's (it makes a difference in my architecture). It would save me some
> > work..
> >
> > Am I correct in making the assumption that I only have to support
> > 257 channels? Will this affect interoperability?
> >
> > TIA
> >
> > -Gabe
> >
> > _______________________________________________
> > BXXPwg mailing list
> > BXXPwg@lists.invisible.net
> > http://lists.invisible.net/mailman/listinfo/bxxpwg
> >
> 
> 
> _______________________________________________
> BXXPwg mailing list
> BXXPwg@lists.invisible.net
> http://lists.invisible.net/mailman/listinfo/bxxpwg