QCon SF: Sanjiva Weerawarana -- WS-* vs. REST: Mashing up the Truth from Facts, Myths and Lies
November 8, 2007
These are my unedited notes from Sanjiva Weerawarana’s talk “WS-* vs. REST: Mashing up the Truth from Facts, Myths and Lies”.
- Sanjiva claiming he’s unbiased, and if he’s biased, it’s because I told him to be :-)
- to be loved and hated for creating WSDL 1.1
- started Apache SOAP, WSIF
co-authored lots of other stuff in the WS-* space
history: why were WS created? people were doing XML over HTTP in 1998/1999
- everyone invented their own way to do security, reliability, transactions, … (e.g. RosettaNet, ebXML)
- not good from a middleware provider POV, such as IBM
- looking for a way to do common things
cynical view: RPC between Java and .NET
WS design rationale:
- world is not all about HTTP, XML, XML Schema
- not all interactions are request/response
- composability of features is key (security, reliability, …, needed, but not al the time)
- SOAP 1.1 section 5 created because XSD was not yet a standard
Biggest criticism of SOAP in 2000: lack of security
Question: is the design rationale from 1999 still valid today? Answer: yes, most of them; HTTP not , Relax NG is a lot better than XSD, but bad luck/bad timing; XML is probably no longer questioned
- Stu: WS were very much centered on XSD. Sanjiva: WSDL is the way it is, including the ability to include COBOL types
- WS-* is just overhead unless you have something in your SOAP headers (shows SOAP message without headers)
- Lie: always do SOAP
- if HTTP(S) + XML is enough for the problem, more power to you
- Glen: possible advantage: future extensibbility
- Dan Diephouse: most people are doing this because of WSDL, not of SOAP
- Lie: you never need SOAP
- Sure, let’s all go back to 1998
- REST-* is on its way - ARGH!
- HTTP-R, anyone?
- no commonly accepted, a.k.a. interoperable REST model for message signing, non-repudiation, reliable messaging
- Question: is there an accepted and interoperable standard for WS-RM? A: Yes (with some complaints from the audience)
- Question: What is the reference platform? Answer: Microsoft - they led the path, not IBM
- Lie: WS-* is complex
- reference to Tim Bray’s slide
- reference to our poster
- is WS-* really complex?
- for the middleware implementor - yes quite
- for the app developer: no
- WS-* programmers need to understand XML, XML Schema, WSDL and WS-Policy - if they tell you otherwise, find better software
- WS tool vendors have not given the right software to developers yet, but are getting closer
- Patrick Logan: the last time I tried this, just the basic stuff didn’t work
- Sanjiva: most of the problems come from databinding
- Patrick: it was the simple stuff that didn’t work, that is my definition of complex
- Sanjiva: let’s talk about this and we’ll sell you some consulting :-)
- Stu: complexity is because of differences e.g. in document/lit/wrapped vs. other styles, other example: WS-Attachment problems because of one team mis-understanding the MIME spec. Specs are getting better, but they may be to complex
- analogy: is TCP/IP complex: for the stack implementer, yes - for the app developer: no
- HTTP has had 15 years to become as solid as it is (myself: exactly! that’s why we should use it)
- is HTTP complex? yes - see Sam Ruby’s ETech 2005 presentations
- Patrick: enormous pressure by the vendors on organizations, but products not ready
- Lie: SOAP is about RPC
Reality: 0.9 RPC/HTTP only, 1.1. RPC, not HTTP only, 1.2: messaging format with RPC supported
Lie: REST is easy to learn
- Large list: if you look at all the standards, including lots of the RFCs
- Stu: WS-* includes all of these
- Old testament: Roy’s dissertation
- New testament: RESTFul Web Services book
- REST is simple
- not true: increasing abstraction, comparison to UML/MOF (MDA was supposed to rule, and now is dead)
- Stu: not true,
- True REST is still an art form
- Criticising AtomPub took a lot of effort by a lot of really smart people, took 1.5 years - points to Web3S as showing that even they didn’t get it right
- Are you smart enough to build a RESTful application? he claims he is not
- are average developers & architects able to design RESTful systems correctly
- Question: why is everybody implementing WS-*? Answer: because it’s the first time everybody agreed
- Disagreement from the audience, no it’s the second time - see HTTP. Again: no, it’s the third time: TCP/IP
- Comment from the audience: Even Flickr couldn’t work without REST (because of caching)
- Hypermedia as the engine of application state not for normal peoples
- Microsoft personae: Einsteins, Morts and Elvises
- Comment from the audience: it’s the phrasing, not the content
- my claim: if the phrase were “use documents with links” it would capture 80% of it
Lots of subtleties - method safety, idempotency, caching, uniform
Lie: REST doesn’t need a description language
- Truth: you need to have tooling, thus you need a description language
Stu: the real debate is about whether we should have a generic description language or specific MIME types
Lie: content negotiation, the savior
- reference to Larry Masinter posting from 2006
Comment from the audience: if you write a consumer, the story changes
Myth: REST programmers eat the payload directly
- Reality: WS- tools have made programmers lazy
- Story of Axis2: they wanted to get rid of data binding, but programers didn’t accept it
- Programmign XML in Java still sucks
- RESTfulness won’t remove programmers’ urge to look for restfulness
- other languages have better XML supoprt
ES4 will drop E4X?
Myth: WSDL is widely accepted and used
Reality: See you at QCon 2015
Myth: WSDL can’t describe RESTful services
- Reality: WSDL 2.0 can describe any RESTful service
WSDL 2.0 and WADL are basically the same, inverted in thinking
Myth: HTTP, the one true protocol
- Enterprisey: JMS, SMTP, TCP, IIOP, MQSeries
Cool: Jabber/XMPP, YahooIM, SIP
HTTP’s uniform interface is the greated - untul we need just a tad more
- WebDAV and DeltaV: a whole bunch more
- PATCH: just one more to get it right
Stu: There’s nothing wrong with having an extensible uniform interface
Myth: REST is multiprotocol
in reality: was one of those nice in theory, never been done in practice
true, very true: as long as you don’t want security
Discussion with Pete and Dan: no, not true, client
Myth: SOA was a response to REST
not true,
REST is HOT, WS-* is NOT
REST is at the top of the hype curve, WS-*
Damned Lie: It’s easy with REST or WS-*
Distributed computing is hard, no matter what
Sanjiva’s Advice: don’t get caught up in the hype
- REST and WS-* both have strengths
- best offer both using tooling, similarly to what POJOs did to J2EE
- building scalable interoperable systems is still hard
- The, on the next slide: Naaah. Just switch to Erlang. :-)
I had a lot of disagreements with his statements, but couldn’t have wished for a better talk in this track.
About
This page contains a single entry from Stefan Tilkov's Random Stuff posted on November 8, 2007 5:09 PM. The previous post in this blog was W-JAX Sessions. The next post in this blog is QCon SF: Steve Vinoski -- REST Eye for the SOA Guy. Many more can be found on the main index page or by looking through the archives.