httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Kosut <ako...@organic.com>
Subject Netscape... arrgh (again)
Date Thu, 25 Jul 1996 08:46:29 GMT
*sigh* Someday I'll stop expecting Netscape to get things right.

Anyhow, I'm working on implementing HTTP/1.1 in Apache, getting byterange
support working, and I decide to go check to see how Netscape does it. So
I punch up a Netscape server I know has byterange support, from previous
tests, help.netscape.com (running Netscape-Enterprise/2.0a). I enter a
couple requests, see what happens. Seems to work okay, as advertised.
Doesn't support If-Range, but that's okay, I suppose.

Then I notice something I never noticed before. When I request a multiple
byterange, the content type is tagged as "multipart/x-byteranges". I stare
at this a minute, then decide that might be okay. It's not the greatest
idea in the world, but heck, it's an alpha server (2.0a), was probably
written when the spec was being revised a lot, they decided to adhere to
the MIME "standard" of using x- prefixes (never mind that HTTP doesn't do
this). I decide this isn't an *entirely* bad thing. Plus I remember (and
check) that Adobe's byterange CGI script
(http://www.adobe.com/cgi-bin/byterange3) does the same thing. Okay, I
figure, it's experimental.

So, I set off to find a released version of Netscape 2.0. I find a server
running Fastrack 2.0 (wwwus.netscape.com). Hmm. It does the same thing,
multipart/x-byteranges. Guess they never changed it. I sigh, but am still
not too worried. The HTTP/1.1 spec isn't released, they're still an
HTTP/1.0 server. It'll make life a little rough on clients, but that's
okay, I suppose.

Then I happen across a Netscape Enterprise 2.0b4 server
(home.netscape.com). I figure it's probably going to do the same thing. It
doesn't. Instead, what it gives is "multipart/mixed". This causes me to
swear out loud, and start to write this email message. *sigh* What the
heck do they think they're doing? Making everyone's lives miserable,
that's what. Especially anyone who tries to write a browser to the
HTTP spec and expects it to work.

Navigator probably never even sends multipart requests, so heck, why
should they bother to get it right in their server? Heck, even if
Navigator did send them, it probably wouldn't even check the content type.
Reminds me of Navigator's wonderful authentication paradigm; it only
supports Basic auth, but doesn't bother to check if the WWW-Authenticate
header actually says "Basic" in it. It just looks for "realm=", and
prompts you. Better yet, even if it doesn't find it. So if you've
got a resource protected with digest authentication, Netscape makes
pretend that it can access it. Took me several really puzzled minutes when
I wrote mod_digest before I figured out Netscape *didn't* support digest
auth. I sent Netscape a bug report, but they never fixed it. Another
example: Go to any of the *.msn.com sites. They send "WWW-Authenticate:
MSN", presumably indicating to IE or whichever "I want to authenticate on
the user's MSN account". Which makes sense. But Netscape just pops right
up with "Please enter username for "unknown"". Arrrgh.

Enough already. I'm going to try and get some sleep. When I wake up, I'm
going to see about having Mountain View's zoning people
"accidentally" levelling all the buildings on East Middlefield Road.

-- Alexei Kosut <akosut@organic.com>            The Apache HTTP Server 
   http://www.nueva.pvt.k12.ca.us/~akosut/      http://www.apache.org/


Mime
View raw message