2008/11/3 Mark Baker <[EMAIL PROTECTED]>: > On Mon, Nov 3, 2008 at 6:47 AM, Steve Jones <[EMAIL PROTECTED]> wrote: >> 2008/11/3 Mark Baker <[EMAIL PROTECTED]>: >>> On Sun, Nov 2, 2008 at 4:29 PM, Steve Jones <[EMAIL PROTECTED]> >>> wrote: >>>> 2008/11/2 Mark Baker <[EMAIL PROTECTED]>: >>>>> On Sat, Nov 1, 2008 at 2:28 PM, Steve Jones <[EMAIL PROTECTED]> >>>>> wrote: >>>>>> "millions of developers" are obeying the hypermedia constraint? I'm >>>>>> assuming that you have no figures to back up that bold assertion. >>>>> >>>>> The only figure I need is this one; >>>>> >>>>> http://www.cs.uiowa.edu/~asignori/web-size/ >>>>> >>>>> plus the knowledge that pervasive use of hypermedia is a necessary >>>>> condition for the Web to have reached that size. >>>> >>>> Errr Mark, you do know the difference between the use of hypermedia >>>> and the concept of hypermedia as the engine of state don't you? The >>>> two are not a single thing, I'm surprised you aren't aware of that. >>> >>> Of course I'm aware of the difference, Steve. But it's immaterial >>> when the links contain sufficient information for the next state in >>> the application state machine to be reached through the reception of a >>> new representation. Like, for example, when they're http links. >> >> This thread started off with an example from Roy Fielding of why >> people using HTTP and claiming it as REST are muppets and should stop >> doing it. You now appear to be saying that as long as you use an HTTP >> link then its fine to break every REST rule in the book so RPC is fine >> as long as you do it via HTTP links. > > What are you talking about? I think the discussion is quite clear, > and that nobody could reasonably make that assumption based on what > I've said in previous messages.
I completely agree based on previous messages. But a couple of times (here and the "protocol" one) you've pushed the idea of HTTP adoption as being indicative of REST adoption. The two things are different. As was said before WS-* uses HTTP and it isn't REST. Your references are about purely websites, this has NOTHING to do with proving your statement on REST adoption being in the "millions". > >>>>> And if you're looking for examples or counter-examples, don't bother >>>>> looking at how individual sites are structured, because that's not >>>>> what yields a "Web". Instead, look at inter-site integration. >>>> >>>> Oh for pities sake, so show me the millions of people who have done >>>> inter-site integration then. >>> >>> Would one suffice? I just visited your blog, and noticed it presented >>> a link to your book's page at InfoQ.com. Well done! >> >> Oh boy. You really do like clutching at straws in your strawman. >> >> Hypertext linking is NOT the same as Hypermedia as the engine of state. >> >> As an example from my blog there is NOTHING that says what the purpose >> of the link is or what the type of the destination page is or indeed >> what the reason for the link is, the best you can say is that the link >> has a description but then you can't use that for inference as its >> pure untyped English. > > None of what you describe there is necessary for inter-site > integration using the Web. All that's required is a link, and that's > what you've got. Take a bow already, and stop being so modest. 8-) This only works if you define integration as "linking two things" rather than having anything that is actually useful for computers to use for integration. Link traversal isn't integration in the same way as FK traversal isn't integration in a database. Just a quick question: Are you saying that HTTP and the href tag is sufficent for a site to claim it is doing REST style integration? This way it would be the case that sites that just use POST for everything would be fully REST compliant. > >> I thought you understood about REST and the difference between people >> who just use HTTP and claim REST (what Roy was moaning about) and >> people who are actually doing REST. Clearly you have a massive >> difference of opinion with Roy on the minimum criteria to define REST >> compliance, I look forward to your blog post explaining why all that >> is required to "do REST" is a hyperlink. > > No, I don't Steve. You just don't understand what I'm saying. That's > ok, it happens, but you might want to tone down the incredulity a tad. Ahhh its me. Thought it might be, because you seemed to be saying that every website was proof of REST in action, which it quite plainly isn't. My incredulity stems from the idea that anyone could list up stats like protocol usage and number of websites as evidence of the number of developers using REST. It is like people saying that the number of cows in a state indicates the number of voters. Mark, I'm happy for you that REST is being used more these days than it was 2 years ago, but don't start becoming a historical revisionist and claiming that the whole Web is REST. Fielding's well made point is that people can use HTTP and fundamentally not use REST. Steve > > Mark. >
