Lies, damned lies, and bloat stories.
shadowgate:/usr/lib/netscape$ du -sm
shadowgate:/usr/mozilla$ du -sm
Ahh... Mozilla is so &amp;amp;amp;amp;quot;bloated&amp;amp;amp;amp;quot; that it's 13 megs *smaller* than NS 4 WITH debugging, IRC client, etc! Oh my!
Maybe you should take a look at a finished product someday. Comparing 0.7 of a project of this magnitude with 5.5 of another project is unrealistic at best. IE 1, 2, and 3 were rather .. rough. IE 4 and 5 for the x86 still have a lot of weird bugs that can be tickled if you nest tables properly (like on my homepage, which only misrenders in IE, yet is totally standards compliant).
The last time I had to install IE 4 on a Win95 system, it required 100 megs to install. How bloated is IE 4?
Maybe you're basing your 'bloat' ideas on the fact that it loads slowly, and still has memory leaks. I'd expect that in a development version. I run development versions of programs on my workstation occasionaly, when testing things. They're not meant for general consumption! And if there are projects out there that are stable and useable in pre-1.0 form, it's probably because the developer isn't comfortable releasing a 1.0 without some additional documentation/features/etc.
If it's such a problem, I'm sure you could write a nice XHTML browser from scratch fairly easily. GdkPixbuf, Gnome Canvas, gnome-xml, all provide you with the tools for the job. I've worked with them myself for my RSS parsing programs. Just be aware that adding CSS to that, and then making it work with legacy SGML-descended HTML is very hard. I applaud the Mozilla group, and feel glad I can use their work.
Bad way of controlling access.
Restricting access via referrer is poor. Why? Because people will fake it to get at what they want. It's not hard. Plus any privacy programs (such as Junkbuster (http://www.waldherr.org/junkbuster/)) spoof the referrer. This leads to problems.
Sites such as Xoom, Angelfire, and Tom's Hardware are all unusable to me (and others) because they enforce "referrer checks" ... If I try to load a story page in Tom's hardware, I just get the "default" front page (and annoying hammer logoes everywhere). The web is about linking, yet this breaks it completely! I can't even point to a specific page of a story, let alone read it. I don't browse Tom's Hardware anymore, and I won't browse other sites which break because of the spoofed referrer.
My suggestion: enforce IP checks (has this IP been on another page in our site in the past few minutes? If Yes, ...). This is very easy to do with a little PHP or Perl code included in your pages, and could probably be written as a mod very easily. Your site would still work for privacy concious individuals, and deep linkers would have a bit more trouble accessing your content.