Well, it's fast. I mean I hope it will be still fast, when not my
simple xslt but the one for arbitrary documents is used. Network
response time can be time consuming too (many hits on server, if that's
a factor, on the other hand, there's the extra bandwidth required to
download the whole file, certainly). People are trained for
multitasking when they use a browser, just look at the tabs. And with
all the flash "Loading", we're pretty much getting used to just wait
for a longer task and doing something else for that time. And in
response, we expect that when our attention is there, the application
really take care for us.
You're certainly right, this can be achieved by the method you explain
too, I can't really say anything against that. Of course There are some
things to think on finding the right amount of pre-catched material,
and how to select it. But it can be solved using the table of contents,
or the backward-forward links as a data source.
Another potential benefit of having the transformation in the browser
should be embeddability. That's not without development demand either,
but I can imagine that the same effect can be reached the way Google
Ads does it: the user inserts some javascript into the webpage, and the
content is there, inside a box, with inner navigation, without ever
leaving the page.
Of course, this can be achieved with the method you descripted too,
given that the html files we're talking about, don't contain anything
just what has to be between the <body></body>tags, and the
table of contents is a convenient enough resource for the javascript to
rely on.
Yours,
Arpad
2010.03.12. 20:09 keltezéssel, Sam Fischmann írta:
This is really interesting and looks pretty cool, but I am
curious what the advantages are of making the browser perform the
transformation, instead of:
Doing a chunked HTML transformation beforehand and putting the
HTML files on a server.
Using javascript to download and cache the HTML files locally
on the client when the user loads your page.
Then you have the advantage of a local cache without the
disadvantage of worrying about transform time, browser XSLT
compatibility, transforming arbitrary documents, and difficulty of
adapting the DocBook stylesheets to work in this special use case. The
plain old HTML linking mechanism would work fine, too, provided that
you hijack click events to perform a lookup in your local cache before
requesting the file from the server.
-Sam
