Hi Sebastian, > What I found originally was that there was a difference between the > output produced by 1.4 and 1.5, in that 1.4 adds an extra space before > the terminating "/>". I'm not sure that it does this for every tag, so > the only way I could see to make sure that the output was the same for > both was to remove the extra spaces. > > It seemed unnecessary to strip the spaces for 1.5, so I made it > conditional. > Later I discovered that some VMs other than Sun added the spaces, so I > changed the check to look for Sun 1.5 only. > > Seems like you have found that the replace needs to be done always. > > The alternative would have been to insist that everyone used the same > JVM and vendor to generate the output ... > > There is another difference between JVMs - tags with multiple > attributes are ordered differently when converted to XHTML. This was > handled by only allowing a single attribute per XML tag, and using > <xsl:attribute ...> for the others. There's a task which checks the > XSL files to enforce this.
I admit it's a long shot, but this sounds like a lot of effort with the intention to make "diff" work on XHTML files. Is there no "xmldiff" tool that knows how to ignore whitspace within tags and the order of attributes? Or is there another reason why these things should remain constant though XML (and therefore XHTML) considers them dont-cares? cheers, Roland --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
