Peter Trudelle wrote:

> Aaron Andersen wrote:
>
>> What, if anything, can those of us who thoroughly hate this feature, 
>> do to get it removed? 
>
> First off, what is the real problem?  Why do you thoroughly hate it?  


Now I wasn't around when all this took place, so correct me if I'm 
wrong, but this is history as I understand it.

Sometime in the spring of 1998 what was then Netscape Communications 
Corporation made the historic decision to release their browser under 
open source, and to create the Mozilla Organization to develop it.  This 
decision was undoubtably discussed in detail amongst the Netscape execs, 
with the pros and cons of doing such an unprecedented thing carefully 
weighed.  Among the pros was the potential to attract outside developers 
to the project, and the possibility of this resulting in better, faster 
development of a better, faster browser.  Among the cons was the fact 
that once the code was under an open source licence the browser would no 
longer be under direct Netscape control; there would be features 
incorporated into Mozilla that Netscape didn't want, and features not 
incorporated into Mozilla that Netscape did want.  This issue was 
partially resolved through Netscape maintaining a separate tree of 
overlays and such that did the things required to make Mozilla into 
Netscape, such as removing IRC and adding AIM.  In the end, it was 
decided to go ahead with the open sourcing plan, mozilla.org was created 
and this all began.

At that time Microsoft had a web browser that was sort of standards 
compliant, but was missing some things, and was full of proprietary, IE 
only tags and extensions, like <marquee> and "document.all".  Netscape 
had a browser that was sort of standards compliant but was missing a lot 
of things, and had a bunch of proprietary, Netscape only tags and 
things, like <layer>.

After a few months of hacking away at Netscapes 4.x code trying to turn 
in into 5.x code, it was decided to totally scrap the 4.x code and start 
over with what would eventually become Netscape 6.x.  Exactly when the 
new Gecko rendering engine was written I do not know, but one thing is 
certain: it was decided that Gecko would support neither the nonstandard 
IE extensions nor the nonstandard Netscape 4.x ones.  By doing so, we 
did not support "document.all".  People would continue to write code 
that used document.all.  People would file bugs complaining that 
document.all didn't work in mozilla.  And every time we would politely 
tell them that we weren't going to support document.all because it is a 
nonstandard proprietary extension, and mark their bugs as WONTFIX.  
Similarly,.we also dropped support for Netscape 4.x's <layer> tag.  
People wrote pages with <layer> in them.  Lots of people.  People even 
had javascripts that detected when a Netscape browser was present and 
delivered pages containing <layer> tags.  And yet we stuck to our 
original goals and WONTFIXed all the bugs asking for <layer> support.

Two years later, on March 16, 2000 someone filed a bug asking mozilla to 
support the nonstandard proprietary extension known as favicon.ico.  The 
first four people to comment on this bug all suggested that mozilla only 
support the standards compliant way of doing this, by using a <link> tag 
or http header to specify that a page had a site icon.  That way, people 
who wanted a site icon could easily add one, but people who didn't 
wouldn't get their servers pounded with requests for a file that didn't 
exist and their errors logs filled up with 404s.  After all, that was 
what we had been doing all along when there was a standards compliant 
way of doing something and an IE way, so why should this by any 
different?  The bug lay dormant for a while, but about a year and a half 
later word got out that a fix was nearing completion and we started 
discussing it on the newsgroups.  In each case, all of the webmasters 
and the majority of the nonwebmasters agreed that automatic favicon 
fetching was not a good idea.  Somehow, over the strong objections of 
many people, it was checked in and turned on in 0.9.7.

When I complain about the number of people who hate this "feature" and 
am told that lost of people like it too.  But lots of people liked 
<layer> and lots of people like document.all.  We don't support them, so 
why this?  Can someone explain to my what makes favicon so special?  The 
only difference I can see is that neither document.all nor <layer> 
wasted the bandwidth or flooded the error logs of innocent webmasters 
who either don't know about them or don't want to use them.  Automatic 
favicon fetching does.  And it's extremely annoying, not standards 
compliant, and just plain wrong.  We shouldn't be doing that. A long 
time ago it was decided that robots would be allowed to automatically 
ask for robots.txt everywhere they went, so that they didn't end up 
where they weren't welcome.  But that doesn't give Microsoft the right 
to just make up new such files at will, nor does it give us the right to 
copy them.

This post is too long already, so I won't even get into the whole "your 
voice as a community member means nothing; if you want any influence you 
have to be a paying customer" philosophy that has surfaced during all of 
this.  But I hope that somewhat answers the question, "Why do you hate 
it so much?" that has been asked a lot lately.  Of course I could go on 
for another few pages if you want me to explain it more.


Aaron Andersen
www.xulplanet.com


Reply via email to