Re: eww/shr: a method for ignoring elements?

[Óscar Fuentes]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Óscar Fuentes <ofv@wanadoo.es> writes:
>
>> Lars Magne Ingebrigtsen <lmi@gnus.org> writes:
>>
>>> See `shr-blocked-images'.
>>
>> Thanks Lars, although that variable affects only images.
>>
>> My primary use case is related to blocking trackers,
>
> What kind?

Things like google-analythics and disqus. When one of those do not
respond, for whatever reason, eww just waits for a long time.

This is part of the message sequence of a web page I just visited:

Contacting host: feeds.nature.com:80
Opening TLS connection to ‘disqus.com’...
Opening TLS connection with ‘gnutls-cli --insecure -p 443 disqus.com’...done
Opening TLS connection to ‘disqus.com’...done
Opening TLS connection to ‘a.disquscdn.com’...
Opening TLS connection with ‘gnutls-cli --insecure -p 443 
a.disquscdn.com’...done
Opening TLS connection to ‘a.disquscdn.com’...done

If disqus.com responds quickly, there is no problem. But if it is slow
or unresponsive (something that happens a few days every other month)
eww hangs on

Opening TLS connection with ‘gnutls-cli --insecure -p 443 disqus.com’...done

At that point pressing C-g makes the same message reappear (it seems
that eww re-tries the operation) which hangs again. Several C-g's are
required to unblock eww (and Emacs). To make things worse, every now and
then Emacs crashes on that sequence of C-g's. And all this for accessing
a site that provides nil or negative value to the page's content.

>> javascript-based commentary sections (disqus, bazaar-voice) and
>> similar stuff.
>
> shr doesn't do JS...

I know, but the sites are contacted anyways (eww has no way to know that
the content being asked for depends on the availability of JS.)

>> Besides privacy control, too often eww hangs because the sites which
>> are contacted by those elements do not respond quickly and it is
>> necessary to press C-g several times to make eww responsive again.
>
> Images should be loaded asynchronously.  The only synch element should
> be the DNS resolution, which we must make asynch one of these days.

Images is not the worst problem (although moving content around for
showing an image after you begun to read the text is more annoying on a
text-based browser than on a GUI-based one, in my experience, and
implementing a poor-man's version of Ad-Block would be nice too), but
the problem is with eww blocking while it contacts stuff mentioned on
the HTML, see above.