Re: Feature request

On Thu, 2005-14-04 at 10:35 -0400, Richard Lister wrote:
> Sridhar Dhanapalan <[email protected]> wrote
> > I used to use Privoxy, but I found that it was slowing me down rather than 
> > speeding me up.

> You could try turning off any filters you have. In order to apply
> filters privoxy will load the whole HTML page first. Though you get
> the final page at the same time, it can *appear* to load slower because
> most browsers render incomplete HTML as they get it. This will
> only be noticeable over a slow connection. Subsequent loads will be
> fast thanks to privoxy caching.

Note that AdBlock (for Firefox or Epiphany) won't cause this delay. On
Firefox there can be a delay (CPU-bound) if you have lots of filters and
are browsing a big page, since each filter will be compiled and run
against each piece of embedded content on the page, in Javascript. In
Epiphany there is a delay, also, but I'm guessing it's an order of
magnitude faster because regular expressions are only compiled once.
(Admittedly, I never benchmarked it; but with the extension enabled, I
notice no delay whatsoever.)

Anyway, with either AdBlock extension, you see the main page as it is
being loaded; there is no delay over a slow connection.

> > plus a CSS file:
> 
> Doesn't this just transfer the filtering burden from privoxy to the
> browser? I don't know enough about gecko/CSS to know how this would
> impact performance.

You cannot use CSS to block content. (Well, unless it was used to load
content in the first place -- if you put "background-image: none !
important;", you could block images which would have been loaded using
the "background-image" property.) All you can do with CSS is *hide*
content.

If you use a CSS file, your bandwidth is still being spent downloading
images, even though you'll never see them. On dial-up, this is an issue.
But the impact on page rendering speed is minor.

I imagine the "perfect" ad-blocking solution would consist of all three
solutions: a proxy (or in the browser, proxy-like filters applied to
pages as they're downloaded), a URL-based blocker such as AdBlock[1],
and some CSS rules to make the sites render nicely after ads are
removed.

[1] I think such a thing would be necessary so the browser would be able
to distinguish between a download which was intentionally blocked and a
download which failed for network-related reasons; a proxy can't provide
that information unless it's integrated in some manner with the browser.

-- 
Adam Hooper <[email protected]>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQBCXojy90GtBxQ61zARAty4AKCGOxHNnGdmdPq7WbDqELuPABW6AACguMTO
VenX6ooU/P5UKMlTn89rH9I=
=6AmE
-----END PGP SIGNATURE-----