On Do, 2004-06-17 at 10:46, John Wilson wrote:
The strengths of XML-RPC are its simplicity and interoperability with a very wide range of other implementations. The creator of the spec and the person who claims ownership of the XMl-RPC trademark has repeatedly and vociferously stated that 'vendor extensions' are unacceptable because of the impact on interoperability.
I very rarely agree with Dave Winer but I do see his point on this.
I would most strongly oppose any introductions of 'vendor extensions' of this sort to the Apache XML-RPC implementation.
I support you 100% in the case, where interoperability matters.
However, there are those of us, who like XML-RPC because of its simplicity and not because of interoperability. And in that case, I can see no point in limitations as long as any violations of the spec are clearly visible in software and network protocol.
Jochen, I defer to John on the timezone interpretation of what's generally accepted as output by common server implementations. However, like usual, I'm of the opinion that servers should be gracious about the types of input accepted, so long as doing so isn't explicitly against the specs (XML-RPC and ISO 8601 apply here), or overly complicates the code. Assuming your patches to parse time zone and to interpret millis follow these rules, send'em over. However, for the sake of interoperability with less forgiving servers, our client will not generate output of this less-common sort.
- Dan
