Loading ...
Sorry, an error occurred while loading the content.

981Re: [soapbuilders] Re: The Interop tests and BDG (was : some questions/observations re: BDG)

Expand Messages
  • Dave Winer
    Apr 1, 2001
    • 0 Attachment
      Andrew, first, it's great to have you be part of this discussion.

      For everyone else, I had a few talks with Andrew before this process
      started.

      One of the principles we agreed to is that the spec comes first.

      You say that if we "ignore" a "number of important features" that it will
      "not be SOAP."

      That's vague.

      Dave


      ----- Original Message -----
      From: "Andrew Layman" <yahoo@...>
      To: <soapbuilders@yahoogroups.com>
      Sent: Sunday, April 01, 2001 8:38 PM
      Subject: Re: [soapbuilders] Re: The Interop tests and BDG (was : some
      questions/observations re: BDG)


      > Thanks. Glen is right. There are a number of features in SOAP that are
      in
      > there for a purpose. Ignoring them has two important consequences:
      >
      > 1. The resulting implementation will not be of SOAP. It will not
      > interoperate with proper SOAP implementations.
      >
      > 2. The application acting in this way will lack the benefits of the
      SOAP
      > feature. (In the case under discussion below, it will not have a good
      > versioning mechanism, and, as many of us know, you cannot easily add a
      > versioning mechanism to v2 of something if it was not provided for in v1.)
      >
      > ----- Original Message -----
      > From: "Glen Daniels" <gdaniels@...>
      > To: <soapbuilders@yahoogroups.com>
      > Sent: Sunday, April 01, 2001 8:15 PM
      > Subject: Re: [soapbuilders] Re: The Interop tests and BDG (was : some
      > questions/observations re: BDG)
      >
      >
      > Paul sez:
      > > Hi, Keith!
      > >
      > > > > When the BDG says to do X, does it mean you don't have to accept
      > > > Y
      > > > > (where Y is perfectly acceptible SOAP, such as a datatype not
      > > > > mentioned in the BDG)?
      > > I think answer should be 'no'.
      >
      > +1
      >
      > > > > Corrolary: Will a BDG implementation correctly fault when sent a
      > > > SOAP
      > > > > Header it doesn't understand?
      > > I think answer is 'no' again. Why? If I send
      > > <SOAP-ENV:Header><my:something xmlns:my="..."/></SOAP-ENV:Header> to
      > > MS tollkit will it fail request? I don't think so, unless I specify
      > > mustUnderstand attribute (and even then only if mustUnderstand is
      > > supported). The same thing shoudl be true about BDG. For me it's just
      > > common denominator, otherwise how will you know that something you
      > > need it supported on other side without testing it first? I can send
      > > message with multiple references and there is no way to tell
      > > beforehand, will server handle this request or not. Now imagine that
      > > server has no knowledge about id/refs. Will this server fail such
      > > request? 99% that answer is no, because it'll just count IDs as
      > > another attribute, so implementation will get some STRANGE values,
      > > but occasionally they could be perfectly valid for it. That's why I
      > > tried to cover as wide field as possible in my implementation to
      > > avoid surprises of such kind.
      > >
      > > Speaking about BDG, it's trying to define minimal subset of
      > > requirements. Consider practical point. Userland implemented BDG and
      > > stoped their development. Will they be able to interoperate? yes.
      > > Will they fail request with mustUnderstand? probably no. Is it
      > > important for their case? probably not. That's limitation of their
      > > implementation. Why whould you send Header with mustUnderstand
      > > without having proper contract with them? Big question is will OTHER
      > > stop at this point? I don't think so. I don't even think that Dave
      > > will :). imho :)
      >
      > Paul, I strongly disagree with this idea. The whole XML Protocol / Web
      > Service philosophy is about systems interoperating in a fairly loosely
      > coupled fashion. One of the exciting things that SOAP buys you in this
      > regard is orthogonal extensibility - i.e. headers which let me throw
      > arbitrary extensions into our communication without your needing to know
      > about them up front. The mustUnderstand attribute, which lets me tell you
      > what extensions are really critical in my (the sender's) eyes, is a
      REQUIRED
      > part of the SOAP spec, and should, IMHO, be supported at least enough to
      > return a fault if you get a mustUnderstand header and you a) don't grok
      it,
      > or b) don't process headers at all. I just recently changed the Apache
      > implementation to do this, and it's pretty trivial - if you don't do
      > headers, just look for anything in the header element with
      > mustUnderstand="1", and fault if you see them.
      >
      > I would like to see this get added to the BDG (which I can't seem to get
      to
      > at the moment, but I assume it's not in there as yet...), actually. If
      this
      > sort of thing (i.e. MUSTs in the SOAP spec) can be dropped, I have some
      > major worries about obtaining any sort of true interoperability as web
      > services start to really ramp up and use this stuff.
      >
      > --Glen
      >
      >
      >
      >
      > To unsubscribe from this group, send an email to:
      > soapbuilders-unsubscribe@yahoogroups.com
      >
      >
      >
      > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
      >
      >
      >
      >
      >
      >
      > To unsubscribe from this group, send an email to:
      > soapbuilders-unsubscribe@yahoogroups.com
      >
      >
      >
      > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
      >
      >
    • Show all 48 messages in this topic