So, to commemorate
the passing of Netscape,
I installed a copy of Netscape.
Specifically, version 0.9b.
I installed it under the Windows emulator in Ubuntu, partly because
I just didn't feel like booting my MSWindows laptop, and partly
because I didn't want to find out what hilarious and possibly viral
effects this fourteen-year-old installation binary of mine would
produce in a non-virtual MSWindows setup. In any case,
the emulation ran flawlessly.
First, I want to say that it's very very strange to be alt-tabbing
and to see this:
* * *
Now, here we go, starting with a blank screen:
To take you to the browser's homepage, you can click the M logo
(for Mosaic Communications Corporation, before there was that
legal trouble and they had to change it to Netscape Communications
Corporation). But that homepage is 404. You can hit any of
those "Guided Tour" etc buttons, but they are 404. In fact, they
all take you to mcom.com addresses that all redirect
[Big ol' edit: No, they all work
now! Because...]
and emit a 404
page which, incidentally, has an HTTP header of "Content-Type:
text/html; charset=UTF-8". This early version of Netscape
doesn't know how to parse ";..." constructs, so it thinks that
"text/html; charset=UTF-8" is one big MIME type string, and so
it says:
It says that a lot now, because apparently most servers
are configured to emit that as their default HTML charset
setting. Similarly, it can't even make sense of "text/html;
charset=ISO-8859-1" for the Content-Type, which is ironic, because
ISO-8859-1 is the only encoding that NS 0.9b can actually understand.
Moreover, NS 0.9b is like many early browsers in that it doesn't
send a "Host: ..." header (lacking in HTTP 1.0, but now mandatory
in HTTP 1.1), so that makes accessing most current web sites even
harder. So I can't access Amazon, Google, or Wikipedia.
IMDB pages actually load, if you hit Stop at strategic times. Still,
the crashing; and also inexplicable DNS failures and network hangs.
Between the Host problem and the Content-Type problem, I can't
even look at enough web pages such that I could make this NS 0.9b
crash as much as I remember it being prone to. Or maybe it's
just that Wine is more forgiving of wonky system calls than MSWindows
3.11 and/or Winsock were.
(Obviously it has no idea about " ", and since it has
never heard of a <style> </style> element, it
skipped over the tags and revealed the content.
It occurred to me to think of what web server would happily
speak HTTP 1.0, requiring no "Host:", and emitting no adventurous
Content-type type. Then I remembered— it was one
I had written myself, podwebserver! And it basically works!
Here it is, browsing The RTF Cookbook:
Despite its real problems communicating with most web servers,
it renders local files just fine:
And there's an interface for directory browsing
(here hinting at the fact that this program only understands
"8.3"-filenames.)
Now, when I said that those nav buttons under the "Location" bar
don't work, I lied: one works, the Newsgroups button.
Hitting it the first time says that it sees no newsrc for you, so it
makes one, starting out with you subscribed to these groups:
Hitting any of those brings up a whole mostly-working interface for
reading newsgroup topics...
and messages...
The elaborateness of the news interface contrasts with how
rudimentary the mail interface is:
Pick Help : About Netscape and it brings you to
"about:", which shows this page:
And if you know to key in the location "about:authors",
you get:
But enough with the Sartre.
And that's it for Netscape 0.93b, from 1994.
Now it's fourteen years later, and to post this message, I'm using
Firefox, basically a distant descendant of Netscape.
Notice the 'Server Development' guy, 'Rob McCool'. Now, look in most any of the apache web server source codes to see: 'Code originally by Rob McCool' to get an inkling of the impact of this little group.
Sure, there are many non-mozilla derived browsers in the wild out there nowadays, but if you add up mozilla-derived source *and* apache-derived servers, this team's code probably still serves up a substantial percentage of teh Interwebs, even today.
(I used to explain
"a patchy server", but then there's
the question of what a program code-patches are,
and so on. I should just say
"it was patched together from other stuff".
Or just say
"it's a long story,
something to do with helicopters, you don't want to know".)
since it has never heard of a <style> </style> element, it skipped over the tags and revealed the content.
Which is why styles should always be commented out. I was disappointed that commenting out header styles wasn't made a mandatory part of the spec (it's mentioned, but since the comments are dependent on the style language, it's nothing more than a suggestion). This hits me more than most, because my MUA is HTML-aware, but not style-aware, so I get to see endless pages of mso-random-crap: true; mso-more-unnecessary-wankery: 15; mso-non-standard-selector: naturally; at the start of many messages...
I use exmh. Yeah, it's showing its age now, but I've been using it for so long, and it still does what I want, which none of the other MUAs I've tried seem to manage, so I'm sticking with it for now.
Both XML and SGML parsers (well, 'SGML parser' isn't a well-defined term, but you get the point) are free to drop comments before passing up the parsed data, so at least for XHTML commenting isn't necessarily that sensible.
Ahhh! This is so cool. Thanks for posting it. I, too, read about Netscape's recent demise. If only it had been IE that was taken out behind the barn and shot.
2008-01-02 (Wednesday)
Response