----- Original Message ----- From: "Paul Libbrecht" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Thursday, March 14, 2002 10:41 PM Subject: Re: Patch suggestions
> John, > > There's one thing that I am wondering about. > In the description of the MinML project, you seem to say that your major > interest in writing MinML was XML-RPC, right ? > Why not then write a dedicated parser for XML-RPC ?? MinML preceeded MinML-RPC by quite a time. My reason for writing MinML was to set an *upper* limit on the size of an XML parser for the minimal XML subset proposed by Don Park in March 2000. Subsiquently I added more of the XML spec to see what impact this had on the parser size (suprisingly little as it happens). It turns out that MinML is quite fast. This was never a design goal and I'm still a little puzzed as to why MinML seems to be faster than other, larger, parsers. However three XML-RPC impentations (includinng, unsurprisingly, my own) use MinML as their standard parser. If you take Dave's statement that "XML-RPC is XML" (see my previous post) at face value. Then there is very little difference between an XML-RPC specific parser and a full non validating XML parser (attributes and mixed content can be omitted - they don't really add much to the parser size). > > The specification of Dave Winer is known to be frozen so there is no > danger. > And it would bring a big speed-up I believe as, for example, the amount > of attributes is extremely small. > This speed-up would be an interesting advantage compared to SOAP. I would expect XML-RPC to be quicker than SOAP in any case. I'm not sure that the parser speed is the controlling factor in the performance of XML-RPC. Things like the need to provide a correct Content-Length header add significant overhead. John Wilson The Wilson Partnership http://www.wilson.co.uk
