Jeff Stuart wrote:
> Jeffrey Baker wrote:
> > What he said. I don't see what sucks so much about burning a few bytes
> > of bandwith to have your site's big fat icon on the user's desktop! And
> > it isn't like they made up an HTTP ICON request type, they are using
> > standard methods to get the file.
> >
> > -jwb
> >
> Well, let's see. Say the icon is 2K in size. Then, let's say you get 1Mil in
> hits a day. 50% of those are IE5. Ok, that's 500K hits from IE5 asking for
> that 2K file. So that's 1Mb of data a day JUST for that damn icon. Over a
> month is 30 Mb of bandwidth JUST for that damn icon. Bandwidth is not cheap
> folks! Someone has to pay for bandwidth. And some plans only allow for 30 Mb
> a month.
I was unable to glean a persuasive argument from your message, despite
your liberal use of capital letters and explitives. You have several
things wrong. The icon files are 16x16x8 bits, and the format requires
1KB. The icon is not requested on each request, only when a user
bookmarks your site using IE 5. Clearly this number will be a tiny
fraction of all requests a site serves. With caching effects, you can
expect to transmit even fewer bites to the large ISPs. Last, nobody is
forcing you to transmit the icon. You can simply redirect it using
mod_rewrite or intercept it with a mod_perl log handler and discard the
log entry.
> Remember folks, bandwidth costs money! Someone has to pay for it. Maybe not
> you but someone does. EVEN THE FREE HOSTS!
This is not the point. If you have GIFs on your page, you are paying to
transmit those, too. If you don't want to transmit the icon, don't
provide the icon! There are technological means to handle this on the
server side and it takes less time than complaining about it in public.
I think the prevailing philosophy on this subject is dead wrong. On the
web, any and all requests can and will come in. You cannot trust the
client! If you rely on the web, you must be prepared to gracefully
handle all situations that might arise. This might include a broken
client making bad requests to your server in a tight loop (Mozilla M8),
or it might include a client occasionally requesting a file that doesn't
exist (IE 5). Think about this: what if your site gets linked on
Slashdot, and they spell the link incorrectly. Are you going to
complain about all the 404 errors in your log, or are you going to use a
simple RewriteRule to give people the information they are seeking?
-jwb
