Syndication and Blogging - Today and Tomorrow
( Fair warning -- this is going to be completely stream of conscious note taking. I may try to summarize later, but it ain't gonna happen soon :)Check out Tim's, er, 'presentation' at his site.
Tim notes some stats on blog growth, and notes the exponential growth. He is currently looking into real-time, ad-hoc search of the blog space.
Blogs are terribly sensitive to current events (iraq, beheadings, john stewart v. crossfire, etc).
Blogs are competitive with BIG MEDIA (shows graph of inbound links).
Corporate bloggers are growing - Microsoft and Sun being the two largest, respectively.
Tim sez that Sun and MS biew corpblogging as a terribly fantastic thing to do.
PSA for the ATOM working group.
+ Multiple options for embedding content (xml, escaped, base64'd binary)
+ Xml Namespaces
+ CRUD for xml chunks
Two APIs for ATOM
- Rome (Java)
- Universal Feed Parser (Python)
RSS - "world's biggest xml app"
Has been proposed before
ICE - Information and Content Exchange
Extremely comprehensive
Little real impact
News industry has News industry text format (NITF)
- infinitely flexible, extensible, hence interop problems
NewsML - got a little bit of traction, but still shy of RSS's traction
CDF - First "new economy" sponsored xml syndication format
"Grown up way to do Pointcast"
"Not widely deployed"
"Reasonably well-thought out"
"Pre-stable xml"
(I have, at this point, had too much coffee, and not enough break time)
RSS - short, not well written, holes and ambiguities
So, why did it succeed?
Arguments as to why RSS won and others didn't
Gordon Bell - works for MSR
- a time and a place for standards
- defies summarization
Money Quotes
- "no product, technology, or standard before its time"
- standards should be based on real experience rather than committee design
Key XML specification design guideline - MPRDV - "Minimum progress required to declare victory" - Pareto
XML - 3/4 completed before it really got noticed (compare also to SOAP, HTTP) contrast to WS-* specifications
Pulls up the c2 wiki YAGNI specification
Gall's law - (paraphrased) You can't build a complex system from scratch
When people are complaining that a standard is "too simple," it often indicates that it is going to be wildly successful :)
WS-Eventing
- missing the boat ala Gall's law according to Tim
- he vomits over the separating endpoints for managing from subscriber and source
- filtering via a new mechanism rather than leveraging Uri
- subscriptions expire, but also "end subscription" message
Atom - each new version gets successfully smaller
Current internet massive scale leverages 'store and forward'
RSS has two distinct populations - publishers and subscribers
rather than store and forward, uses push then pull
"Let me bash ws-eventing one more time"
There are no normative [schemas] for RSS or ATOM
DTDs are passed their 'sell by date'
Relax NG - not widely deployed
XSD SUCKS
One of the foundational pieces of the GXA is a "Big stinking turd"
No built in data model, object model, or api -- there is a message exchange
Don asks, "Don't you find it funny that you need an api for your stuff RSS and ATOM but you don't for ours"
Tim replies "I don't use an api, I just use a [garbled] parser"
Tim gets nervous when people want to take away the data and hide it behind an interface
"If O/R Mapping is CS's Vietnam, then why does anyone think that XML/Object, etc is going to be any simpler"
Comments