From: Matti Aarnio (tai.gozwxljjnd@wh-stuttgart.de)
Date: Mon Apr 03 2000 - 13:00:22 EEST
On Sun, Apr 02, 2000 at 08:52:48PM -0700, Shawn T. Rutledge wrote:
> > - Instead of running TCP at the mobile appliance, they use UDP
> > to the WAP-GW (I am not sure of this, though), possibly to
> > be able to ignore all the myriad rules about TCP backoffs, etc.
> > (And not to require some 60 kB code for TCP implementation..)
>
> So do they have their own error detection/correction scheme, or just
> let the data be missing if it doesn't make it?
The whole retransmission machinery present in TCP is done
by the WAP stack over UDP frames.
> > - Instead of a general purpose compression schemes, they use
> > XML/WML specific tokenization system which becomes suboptimal
> > the instant the WML gets new attributes.
>
> Hmmm. How is it tokenized? Common tagnames or attribute names
> get replaced by a byte or two?
Something like that.
Worse, the list of tokens (tags, attributes) to compact
is fixed by the WAP specification, but still the each
card-deck contains the dictionary of translations.
> I've been wondering what would be a good replacement for the ad-hoc
> APRS "standard". XML has been proposed as a possible translation of
> the data but noone seriously considers it a good replacement for the
> on-air protocol because of the verbosity and the bandwidth shortage
> at 1200 baud. Maybe tokenized XML would work.
Maybe tokenized XML, maybe not. Often my primary answer
for this type of things is ASN.1/BER -- carrying token
translations might not be necessary.
Say, like sending SNMP traps on a multicast address.
> > unit, but larger scale things like digital mobile phones tend
> > to have terrible time at getting thru a single 500 byte UDP
>
> Why? The voice stream is data too, and it seems to be reliable enough.
The voice stream encoding is such that it is very tolerant of
loosing every second packet (each carrying up to 160 octets of
data in GSM) If you loose/corrupt even single subframe of IP
packet, you loose it all.
> > When a message reaches environment where link connectivity
> > does not (in real life) allow direct end-to-end interactive
> > connectivity, nothing really prevents one from gatewaying
> > from general Internet rules/behaviour to a hop-by-hop routing
> > with e.g. static/periodically revised routing tables a'la UUCP,
> > or BITNET.
>
> Sure but there's nothing easy about it is there? How much can you
> automate it, so that the user doesn't change his habits (destination
> is still terhi.victor@logonet.com) and the sysadmins between the source and
> destination don't have to do anything either? If there's a way I'd
> like to know.
Quite easily. All local special-connectivity machines replicate
among themselves network topology info, and when destination
is to one of them, the transport uses e.g. AX25.
When the message is destined outside, a default-route points
to gateway host, or to neighbourgh which is along the path
towards the gateway host.
> > ...
> > This is at least 6 years old idea, original inventors of HTTP
> > protocol had in mind that we need (and we do!) a way to identify
> > resources (e.g. HTML pages) independent of their locations.
>
> Well there's also "tumbler" addressing as used in the Xanadu project.
> I'd like to see that idea combined with this freenet thing somehow.
> Anyway it seems the really grandiose ideas take forever to implement,
> but freenet has real code running.
The "tumbler" addressing sounds quite much like ASN.1/BER
encoding of certain kind numbers.
> --
> _______ http://www.bigfoot.com/~ecloud
> (_ | |_) dscc.nyofj@dsl-70-212.utaonline.at finger lwawry@upet.ro
/Matti Aarnio <ftmif@balk.dk>
This archive was generated by hypermail 2b29 : Mon Apr 03 2000 - 13:07:22 EEST