On Thu, 2005-04-28 at 15:21 -0500, Curt Arnold wrote: > On Apr 28, 2005, at 1:42 PM, Dave Pawson wrote: > > > On Wed, 2005-04-27 at 15:16 -0500, Curt Arnold wrote: > > > > <snip/> > >>> > >>> Am I making sense? > >>> > >> > >> Okay, I wanted to make sure that you wanted what I think that I would > >> want in that situation. > >> > >> If I were to add that, my initial approach would be to add a new > >> layout > >> (say XSLTLayout) that would use the javax.xml.transform API's (present > >> in JDK 1.4, not sure when introduced). The transform would be fired > >> once for each LoggingEvent. > > Tell me why please? > > To allow you to control the element names, namespaces, position of > information, etc, you would need some sort of template language and > since XSLT is a W3C recommendation, there is a standard javax API, and > an implementation bundled with newer JVM's, it seems like a better > choice than creating a new template language.
Isn't it a properties file? Seems to be to be saying. I want X, I want it as an element content in element P I want Y, I want it as an attribute value on attval Z If XSLT were to be used, then the output would need to contain all possible outputs, and be reduced? I was trying to follow the existing log4j practises > > > > If you are post-processing, you can just use the XMLLayout since all > the information is already there, just maybe not in the form that you > would prefer. I don't recall seeing all information in the DTD? I must admit I haven't used the XMLLayout as yet, but the DTD seemed not to cover everything I've seen in the properties file? > The only justification for a new layout is if we wanted > the initially write the log file as if it had already been transformed. I guess that's where I was going. I.e. specify what I want in the output, including element and attribute names? regards DaveP --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
