Quoting Eduardo Pelegri-Llopart ([EMAIL PROTECTED]):
> I'd be interested in any performace data people may want to share with
> us.
>
> In priciple there should not be any significant overhead (beyond the
> first compile - which can be addressed by precompiling the JSP) in
> using JSP. In practice, the only overhead I've heard is small and it
> is due to the use of the PrintWriter layers. I want to look at that
> more carefully later one, after 1.1.
My comments about overhead were "in theory" and based on the fact that
both JSP and WebMacro rely on similar low level introspection (ie:
beans), which is not free. Probably for the moment WebMacro is slightly
slower since I haven't implemented any optimizations yet, beyond
compiling the template on first access.
The introspection which WebMacro and JSP have to perform, but which
FreeMarker and friends do not, is a small but noticible cost. It buys
you independence from particular object models in exchange for a
slight performance hit.
In theory FreeMarker could therefore be faster than WebMacro or JSP,
except that in FreeMarker there may be some extra object creation
overhead to account for the adapters that have to be implemented--so
maybe it all works out in the wash.
Actually WebMacro does more aggressive introspection than the beans
spec (and therefore than JSP), and is willing to take advantage of
methods like:
request.setHeader("Content-type", "text/html")
which are not easy to get at with only beans introspection. (WebMacro
can view that as a property called "request.Header.Content-type").
Like JSP WebMacro only parses/compiles the template on first access.
Justin
===========================================================================
To unsubscribe: mailto [EMAIL PROTECTED] with body: "signoff JSP-INTEREST".
FAQs on JSP can be found at:
http://java.sun.com/products/jsp/faq.html
http://www.esperanto.org.nz/jsp/jspfaq.html
