Mark Baker wrote: > But FWIW, the uniform interface isn't the only way to get > self-descriptive messages. If the interface is standardized then it > is self-descriptive. That's why SOA/WS (and ONC, DCE, CORBA, DCOM, > RMI et al) - which unapologetically do *not* constrain the interface - > never had, and will never have, self-descriptive messages in the > general case.
I disagree. RMI is uniform. It has a single operation, "invoke". I've stressed this before. Jan seemed to give up on the last go round. I find it very disheartening that so much disinformation is being spread about to try and pump up HTTP as "the only uniform API on the internet" implementation. > It's the main reason we don't see those systems on the Internet. But we do have these systems on the internet. I use them on the internet all the time! It seems that what you consider the internet is "the web". What I consider the internet is TCP/IP networking. The main reason why we don't "see those systems" on the internet is because they don't have to be "seen on the internet" to be useful. Exposing the transport/transfer technolgy behind an API as "the interface" is pretty much the end of extensibility of the implementation. This is why mobile code is so important! > *ALL* Internet based systems have standardized operations. This is > not a coincidence. I think that you are using some pretty broad terms that really don't fit into the mold you are pressing them. The "Internet" is based on TCP/IP, plain and simple. HTTP is but one application protocol that rides on top. It is the protocol that web browsers are using as their primary interface. It's not the only solution, and the browser is not the only useful client. I hear over and over in different places about people moving away from the browser and to fat-clients because of the fact that the browser can not be depended on to "work correctly". There are versioning issues, document implementation issues and simply broken code issues. There's a lot of potential in the browser. However, because there are multiple implementations with multitudes of problems, anything beyond simple text and forms and downloading stuff, is risky at best. The whole GWT business demonstrates how frustrating Google has found it to be. It won't surprise me to see Google creating a new "client" that just downloads and runs GWT code directly. Gregg Wonderly
