[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] XML/XSLT for web templating

---

---

Thanks Rob, and by the way, I had a lot of fun reading and sharing
that thread :)

About your suggestion, and it is is my fault that I did not say that
more explicitly, there's *no* dynamic pages at all.
Pages are just batch-processed and then sent out to a plethora of edge
servers. In other words, there's no handling of http request,
response, session, cookies... nothing at all. Just pure precooked
XHTML and XML files.

Cheers,
Alessandro




On 10/8/07, Robert Koberg <rob@xxxxxxxxxx> wrote:
> On Mon, 2007-10-08 at 10:11 -0400, Alessandro Bologna wrote:
> > 1. Too slow (i proved it wrong already, but still, just yesterday
> > somebody was telling about Freemarker being 15 times faster than XSLT
> > at rendering docbook, and I know that Robert Koberg has something to
> > say about that... :)
>
> Yea, that was a fun exercise... I think I proved clearly that freemarker
> is not 15 times faster. It is indeed slower AND much more complicated.
> Basically, you cannot compare the two. They are different beasts.
> Anybody who claims freemarker is 15 times faster simply does not know
> what they are talking about. And risking a Revulsky infestation, I don't
> think even he claims that anymore.
>
> In freemarker, you have to write a java driver for your transformation.
> In other words, you have to write custom java so the freemarker
> 'transformation' code knows what to do. This has to be written for each
> XML structure/schema you use. If I remember correctly, this was about
> 450 lines for freemarker's handling of a small subset of docbook. Then,
> you have to understand and know the driver's API to write the templates.
> When you use XSL you just write XSL.
>
> It looks like you have a really nice system set up. It is too bad your
> teammates can't or won't see it.
>
> One thing you might consider is pre-generating the pages with sprinkles
> of Velocity/Freemarker thrown in for the truly dynamic runtime
> requirements. This way you can get a kind of caching and let the runtime
> templating library do only as much work as it needs.
>
> best,
> -Rob

---