Re: [Bodington-developers] [Tetraelf-developers] xml foods

[Selwyn L. <sel...@ph...>]

well i think the use case for many clients are

getFeed(complete)
getFeed(delta) [diffs]

and

cacheFeed(complete)
cacheFeed(diffs)
cacheFeed(false)

my concern was you have to read the entire feed to establish the delta 
changes, which is less than efficient

so YES your correct, the idea was that we do need something like...

1. a relationship is established between client and server e.g an 
accessid tied to either a direct login or a login proxy
2. the server establishes the client connecting when the accessid is 
included in the http header [likely POST]
---> insert your http[s] read up to pointer [where accessid=]
3. the server provides the new stuff and only the new stuff unless you 
reset the [proxy / intelligent cache thingy me bob]

i'm fairly confident this would be more efficient when serving large 
amounts and would surely use less bandwidth and i just wondered
if rss / atom were already on to the case

I think the use cases are very close to messaging or publish / subscribe 
but these things are probably less attractive than rss/atom.

Thinking this through from our own perspective is one thing,
but I also think we would see demand from specialist information providers
for
a. feeds for serving up rss/atom in the most efficient way possible
and
b. forcing clients to register for feeds

or maybe I'm just mad :)

Jem Taylor wrote:

>If the page location were
>
>http(s)://server/feedstuffpage?mypermanentidentifier
>
>then you could hold my 'read up to' pointer in a server-side database without needing to know the client - the client has to store and proffer the PUID regardless of the channel used to communicate.
>
>using a cookie to hold the read-up-to pointer gives the same result without needing a PUID anywhere on the server
>
>presumably I'm missing something obvious but it seems to me that the server would have to calculate the content on the fly whatever happens; the question is, should several clients sharing the knowledge of the PUID be able to read the same content, or should they get interleaved slices?
>
>Real World example: If the PUID is my name, and I check every day from different internet cafes as I tour south America, do I get the whole lot each time, or just the latest bit?
>
>
>-Jem (wondering why he reads this list if he never says anything on it)
>
>  
>

-- 
Selwyn Lloyd
Phosphorix 
Open Source Systems for Lifelong Learning
skype: selwyn_lloyd
tel: 07979240124
irc://irc.ionode.org
support channel: #ionode
support email: de...@ph...
web: http://www.phosphorix.co.uk 
forum: http://forum.ionetwork.ac.uk


Phosphorix Limited
Date of Company Registration 15th March 2001
Company Registration Number 4180295
VAT number: 768 9766 42 
Registered Address: REDBRICK HOUSE, SAINT AUGUSTINES YARD, BRISTOL, AVON, BS1 5DS
Offices: Seolea Cottage, Tedburn Road, EXETER EX4 2HD