I have two questions for you guys. Please accept my apologies if they have been asked 100 times already.
I have a system that uses XSLT heavily, including a lot of libraries for building common widgets like <select> and <table>. I'm looking to see what I can do for performance optimization, and am thinking about prototyping XSLTC. QUESTION #1) In my style sheets, I have tags like this: <xsl:import href="rl/Common.xsl"></xsl:import> <xsl:import href="../../proteus/private/proteusxsl/list.xsl"/> How does the XSLTC compiler handle the xsl imports? If I have these common libraries, will it build code for each list.xsl in every stylesheet, or will it reuse a common list.class? Will it be able to handle the import at all? QUESTION #2) In our current implementation, we cache the Xalan transform once we have instanciated it at runtime. Will we see substantial improvement from XSLTC over using this approach? We are currently seeing XSL rendering of substantial amounts of data completing in 60ms, but we are still mostly in proof of concept mode and I want something that will scale when we are running 50x as many transforms. Are people seeing tangible perormance benefits from XSLTC? Any anecdotal evidence you can offer would be much appreciated. Stephen Thompson Wachovia Corporation
