Re: PATCH: Make nnrss include RSS dc:subject

[<yamaoka@jpl.org>]

>>>>> In <m3lkua9ffi.fsf@quimbies.gnus.org>
>>>>>	Lars Magne Ingebrigtsen wrote:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> The advantage of nnshimbun is to remove advertisement.  However, the
>> module to deal with it with each news site is needed (e.g.,
>> sb-bbc.el for BBC), and we need to make major changes in those
>> modules periodically.

> Would it be possible to separate out this into something declarative
> that could be downloaded at will?

This is exactly what the emacs-w3m people were considering.  It
has not been achieved yet, though.  In fact, the number of
modules is too large for a few developers to maintain nowadays.

> The main problem with all these web-centred things is the need to
> update the source code whenever the HTML changes, and I don't see any
> general way to fix that.  To parse Slashdot, for instance, you
> probably have to do that programmatically.

Yes, we will not be able to deal with everything using only
regexps.

> But for stuff like ad removal, couldn't a language be defined for
> that?  For instance a set of regexps (or something) that could be
> applied to the article bodies before displaying (on a per-source
> basis).  The user could update these rules with, say,
> `M-x gnus-update-washing-rules' that would download the rules from
> some central repository.

> The important thing being, of course, that there would be no actual
> code being downloaded, because that would be a security problem.

Or there might be a bug causing a serious accident even if it
was made with no malice.

> Heh.  Imagine rules that would remove Yahoo ads from Yahoogroups
> mailing lists, and Sourceforge ads from those lists, and...

> Gnus ad blocker.  :-)

Perhaps we had better concentrate instead on `Gnus spam blocker'.
:-)