ijbfaq.html

一个不错的服务代理器

you might want to include the word ``junkbuster'' in it
and use the word ``blocklist'' in the file name given in the
<small>URL</small>
so that others can find it with the query given in the previous sentence.
</p>

<h3><a name="zap" href="/cgi-bin/gp?pg=ijbfaq&pr=zap"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
If I see an ad I wish I hadn't, how do I stop it?
</h3>
<p>
If your
<small>ISP</small>
is running the
Internet Junkbuster,
they should have a policy on whether they accept suggestions from
their customers on what to block. Consult their web page.
<p>
<a name="cover">If you are running</a>
the
Internet Junkbuster
yourself, you have complete control over what gets through.
Just add a pattern to cover the offending 
<small>URL</small>
to your blockfile.
Version 1.3 and later automatically rereads the blockfile when it changes,
but if you're running an earlier version you'll
have to
<a href="ijbfaq.html#shutdown">stop it</a>
and restart it.
<p>
<a name="target">To choose a pattern you'll first need to find the</a>
<small>URL</small>
of the ad you want cover.
<p>
<a name="pinpoint">Some people use the</a>
<a href="ijbman.html#debug">debug</a>
<big><kbd>1</kbd></big>
option to display each 
<small>URL</small>
in a window as the request is sent to the server.
It's then usually an easy task to pick the offending 
<small>URL</small>
from the list of recent candidates.
<p>
<a name="source">Alternatively,</a>
you can use
<b><font face="arial, helvetica">
View Document Info</font></b>
(or
<b><font face="arial, helvetica">
View Document Source</font></b>
if your browser doesn't have that).
The
<b><font face="arial, helvetica">
Info</font></b>
feature has the advantage of showing you the full
<small>URL</small>
including the host name,
which may not be specified in the source:
there you might see something like
<big><kbd>SRC="/ads/click_here_or_die.gif"</kbd></big>
indicating only the
<i><dfn>path</dfn></i>.
(The host name is assumed to be the same as the one the page came from.)
<p>
<a name="offsite">But ads often</a>
come from a different site, in which case you
might see something like
<big><kbd>SRC="grabem.n.trackem.com/Ad/Infinitum/SpaceID=1666"</kbd></big>
or longer.
<a name="warehouse">If the company looks like a pure ad warehouse</a>
(as in the last case),
you may want to place just its domain name in the blockfile,
which blocks all 
<small>URL</small>s
from that site.
<p>
<a name="wanted">If the ad comes from a server</a>
that you really want some content from,
you can include enough of the path
to avoid zapping stuff you might want.
In the first example above,
<big><kbd>/ads/</kbd></big>
would seem to be enough.
If you don't include the domain name,
the pattern applies to all sites,
so you don't want such patterns
to be too general:
for example
<big><kbd>/ad</kbd></big>
would block
<big><kbd>/admin/salaries/</kbd></big>
on your company's internal site.
<p>
<a name="image">To speed the blocking of images, some</a>
<small>UNIX <a href="legal.html#not_our_trademark">&#174;</a></small>
users create a
shell script called
<big><kbd>Image:</kbd></big>
containing a line such as
<big><kbd>echo $1 | sed s/http:..// &gt;&gt; $HOME/lib/blockfile</kbd></big>
that adds its argument to the user's blockfile.
Once an offending image has been be found using
<b><font face="arial, helvetica">
View Document Info</font></b>
it's easy to cut-and-paste the line (or part of it) into a shell window.
The same script can be linked to a file called
<big><kbd>Frame:</kbd></big>
to dealing with framed documents,
and
<big><kbd>junkbuster:</kbd></big>
to accept the output of the
<a href="ijbman.html#debug">debug</a>
option.
<p>
<a name="partial">When compiled without the</a>
<i><dfn>regular expressions</dfn></i>
option, the
Internet Junkbuster
uses only very simple (and fast) matching methods.
The pattern
<big><kbd>/banners</kbd></big>
will not stop
<big><kbd>/images/banners/huge.gif</kbd></big>
getting through: you would have to include the pattern
<big><kbd>/images/banners</kbd></big>
or something that matches in full from the left.
<a name="regex">So you can get what you want here,</a>
the matcher understands
<small>POSIX</small>
regular expressions:
you can use
<big><kbd>/*.*/banners</kbd></big>
to block
and any
<small>URL</small>
containing
<big><kbd>/banners</kbd></big>
(even in the middle of the path).
<a name="posix">(In Versions 1.1 through 1.4</a>
they were an option at compile time;
from Version 2.0 they have become the default.)
Regular expressions give you
<a href="http://theory.uwinnipeg.ca/localfiles/infofiles/gcc/rx_toc.html">many more features</a>
than this,
but if you're not already familiar with them you probably won't 
need to know anything beyond the
<big><kbd>/*.*/</kbd></big>
idiom.
If you do, a
<big><kbd>man egrep</kbd></big>
is probably a good starting point).
<p>
<a name="slash">Don't forget the</a>
<big><kbd>/</kbd></big>
(slash)
at the beginning of the path.
If you leave it out the line will be interpreted as a domain name,
so
<big><kbd>ad</kbd></big>
would block all sites from Andorra
(since
<big><kbd>.ad</kbd></big>
is the two-letter
<a href="reference.html#country">country code</a>
for that principality).
<p>
<a name="detail">For a detailed technical description</a>
of how pattern matching is done,
see the
<a href="ijbman.html#o_b">manual.</a>
</p>

<h3><a name="despite" href="/cgi-bin/gp?pg=ijbfaq&pr=despite"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
How come this ad is still getting through anyway?
</h3>
<p>
If the ad had been displayed before you included its
<small>URL</small>
in the blockfile,
it will probably be held in cache for some time,
so it will be displayed without the need for any request to the server.
Using the
<a href="ijbman.html#debug">debug</a>
<big><kbd>1</kbd></big>
option to show each
<small>URL</small>
as it is fetched is a good way to see exactly what is happening.
<p>
<a name="otherwise">If new items seem to be getting through,</a>
check that you are
<a href="ijbfaq.html#show">really running</a>
the proxy with the right blockfile in the options.
Check the blockfile for
<a href="ijbfaq.html#exceptions">exceptions.</a>
<p>
<a name="java">Some sites may have different ways of inserting ads,</a>
such as via
<a href="cookies.html#java">Java.</a>
If you have ideas on how to block new kinds
of junk not currently covered, please
<a href="/cgi-bin/gp?pg=ijbfaq&pr=java">tell us.</a>
</p>

<h3><a name="exceptions" href="/cgi-bin/gp?pg=ijbfaq&pr=exceptions"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
How do I stop it blocking a URL that I actually want?
</h3>
<p>
You can change the patterns so they don't cover it,
or use a simple feature in Version 1.1 and later: a line beginning with a
<big><kbd>~</kbd></big>
character means that a
<small>URL</small>
blocked by previous patterns that matches the rest of
the line is let through.
For example,
the pattern
<big><kbd>/ad</kbd></big>
would block
<big><kbd>/addasite.html</kbd></big>
but not if followed by
<big><kbd>~/addasite</kbd></big>
in the blockfile.
Or suppose you want to see everything that comes from
a site you like, even if it looks like an ad: simply put
<big><kbd>~aSiteYouLike.com</kbd></big>
at the
<em>end</em>
of the blockfile.
(Order is important, because the last matching line wins.)
<p>
<a name="agreed">As well as unblocking</a>
pages that were unintentionally blocked,
this feature is useful for unblocking ads from a specific source.
This might be because you are interested in those particular ones,
or if you have an explicit agreement to accept certain ads,
such as those from a free web-based email provider.
</p>

<h3><a name="children" href="/cgi-bin/gp?pg=ijbfaq&pr=children"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
Can I block sites I don't want my children to see?
</h3>
<p>
Yes, but remember that
<a name="savvy">children who are technically sophisticated enough</a>
to use the browsers' proxy configuration options
could of course bypass any proxy.
This kind of technology can be used as a gentle barrier to remind
or guide the child,
but nobody should expect it to replace the parent's role
in setting and enforcing standards of online behavior for their children.
<p>
<a name="recommend">Some</a>
<small>ISP</small>s
are starting to provide specialized proxies to protect children.
There are two basic approaches: the ``black list'' and the ``white list''
approach.
<a name="negative">The black list approach allows the child</a>
to go anywhere not explicitly prohibited; the white list permits visits
only to sites explicitly designated as acceptable.
<p>
<a name="positive">It's very easy for</a>
anyone to
compile a white list from a page of ``recommended
kids sites'' and to configure an
Internet Junkbuster
to allow access to those sites only.
If you compile with the
<a href="ijbfaq.html#regex">regex</a>
option,
you can place a
<big><kbd>*</kbd></big>
(asterisk) as the first line of the blockfile (which blocks everything),
and then list
<a href="ijbfaq.html#exceptions">exceptions</a>
after that.
Be careful to make the exception sufficiently broad:
for example, using
<big><kbd>~www.uexpress.com/ups/comics/ch/</kbd></big>
as the exception for
<a href="http://www.uexpress.com/ups/comics/ch/"><cite>Calvin and Hobbes</cite></a>
would block some of the graphic elements on the page;
you would probably want a wider exception such as
<big><kbd>~www.uexpress.com/ups/</kbd></big>
to permit them.
<p>
<a name="trust">Version 2.0 has an experimental feature</a>
to permit only sites mentioned in a nominated
<a href="ijbman.html#trustfile">trusted site.</a>
This allows organizations to build lists of sites for kids to browse,
and the software automatically restricts access to those on the list.
<p>
<a name="scan">Many filtering</a>
<a href="links.html#blocking">products</a>
actually scan for keywords in
the text of pages they retrieve
before presenting it,
but
the
Internet Junkbuster
does not do this.
Building a perfectly reliable black list system is hard,
because it's very difficult to state
in advance
exactly
what is obscene or unsuitable.
For more info see our
<a href="links.html#parents">links</a>
page.
</p>

<h3><a name="message" href="/cgi-bin/gp?pg=ijbfaq&pr=message"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
What do I see when a page or graphic is blocked by the proxy?
</h3>
<p>
You usually see a broken image icon,
but it depends on several factors beyond the proxy's control.
If asked for a
<small>URL</small>
matching its blockfile, the proxy returns an
<small>HTML</small>
page containing a message identifying itself
(currently the two words ``Internet Junkbuster'')
with a status 202 (Accepted) instead of the usual 200 (OK).
(Versions 1.X returned an error 404: Forbidden, which caused
strange behavior in some cases.)
Status 202 is described in the
<small>HTTP</small>
<a href="http://www.ics.uci.edu/pub/ietf/http/rfc1945.html#Code202">RFC</a>
as indicating that the request has been accepted but not completed,
and that it might complete successfully in the future
(in our case, if the blockfile were changed).
<p>
<a name="depends">The broken image icon is most common</a>
because the browser is usually expecting a graphic.
But if it was expecting text, or if the page happens to be using certain
<small>HTML</small>
extensions
such as
<big><kbd>layer</kbd></big>
and your browser is a late model from Microsoft,
you may see the words ``Internet Junkbuster'' displayed as a hot link.
<p>
<a name="click">Clicking on the link takes you to an explanation of</a>
the pattern in the blockfile that caused the block,
so that you can edit the blockfile and go back and reload if you really
want to see what was blocked. The explanatory link is generated by
the proxy and is automatically intercepted based on its ending in
<big><kbd>ij-blocked-url</kbd></big>;
even though the site is specified as
<big><kbd>http://internet.junkbuster.com</kbd></big>
no request should actually made to that site.
If one is, it means that the proxy was been removed after it
generated the link.
<p>
<a name="layer">To summarize:</a>
the identifying link to the blocking explanation
is usually turned into a broken image icon,
but it may be displayed on a page alone,
or they may may be restricted to the particular frame, layer or graphic area
specified in the page containing them.
The proxy has no way of knowing the context in which a
<small>URL</small>
will be used and cannot control how the blocking message will be rendered.
</p>

<h3><a name="broken" href="/cgi-bin/gp?pg=ijbfaq&pr=broken"><img border=0 width=14 height=14 src="/images/fb.gif" alt="&lt;Feedback&gt"></a>&#160;
Why not replace blocked banners with something invisible?
</h3>
<p>
<a name="infringe">Many users have suggested to us</a>
that blocked banners should be replaced by a something like a
1x1 transparent
<small>GIF</small>
to make the page would look as if there was nothing ever there.
Apart from making it harder to catch unintended blocking,
this might also displease the owners of the page,
who could argue that such a change constitutes a copyright infringement.
We think that merely failing to allow an included graphic to be accessed
would probably not be considered an infringement:
after all this is what happens when a browser
is configured not to load images automatically.
However, we are
<a href="over.html#notlaw">not</a>
lawyers,
so anyone in doubt should take appropriate advice.
<p>
<a name="done">In a context where the copyright issue