> So, is your concern more about the efficiency of the implementation, 
> alone, or does the extra server bandwidth affect you directly in some 
> way? 


There seems to still be a lot of confusion as to what exactly we are 
objecting to, so I am going to try to use this post to clear up a few 
misconceptions.  This is just my opinion however, so it may not agree 
with every other person who wants automatic favicon fetching turned off.

We do not object to displaying site icons, we object to automatically 
asking a web server for favicon.ico when we have no idea if that file is 
going to exist of not.  This is not a "user interface issue" as some 
have suggested.  It is an issue with how mozilla communicates with web 
servers.  Site icons are really neat and useful.  We know that.  We do 
not want site icons turned completely off.  We like them as much as you 
do.  It is simply our opinion that a web browser should not request a 
file that may or may not exist from a web server unless the user 
specifically asks for it (either by typing the URL in the location bar, 
clicking on a hyperlink, etc.) or it is used in a page that the user has 
asked for.  Not every site on the internet has a favicon.ico file; we 
should not therefor request such a file unless we know there is one to 
request.

The financial issue is non existent, at least for most people.  While I 
do have 2,000 favicon.ico requests in my error log so far this month, I 
do not think it is costing me any money.  My complaint has nothing to do 
with being directly hurt or financially suffering by this.  My complaint 
is based on the fact that I thought that mozilla was supposed to be held 
to high standards, higher so than IE at least.  And yet we have taken a 
completely unacceptable behavior from internet explorer and made it 
worse.  Even microsoft doesn't request favicon.ico at every single web 
site visited, they only do it when you bookmark something.

Some have said this isn't a standards issue.  That is not entirely 
true.  There is a standard, several of them in fact, dealing with how a 
user agent travels across the web and communicates with web servers.  
And one of the rules (although perhaps not written out because it is so 
completely obvious) is that a user agent doesn't go and ask web servers 
for random files just because there might possibly be a file with that 
name that would be useful.  It just isn't right.

Some have suggested that adding link tags to all their pages would be 
difficult.  However, I would guess that 90% of all large websites are 
database driven in one form or another.  In that case adding a link tag 
to every page should be as simple as editing one file.  And if you 
happen to own one of the other 10%, I can send you half a dozen 
utilities that will search through the entire structure of your website 
and replace every instance of "<html><head>" with "<html><head><link 
rel='shortcut icon' href="/favicon.ico>" so you too will be able to add 
link tags to all your pages in five minutes.

I hope that clears up a few things


Aaron Andersen
www.xulplanet.com


Reply via email to