computer question

John Chambers

Message 19430 · 17 Nov 1999 18:00:50 · Fixed-width font · Whole thread

Previous message: Warm-ups (Norah Link)
Next message: Warm-ups (Pia Walker)
Previous in thread: computer question (Sue Wartell ext. 3387)
Next in thread: computer question (Alex Tweedly)

"Mettler, Klaus (LFUST)" <Xxxxx.Xxxxxxx@xxxxx.xxx.xxx.XX> writes:

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

(This mail reader doesn't understand much; it lets me take care of
the "understanding" part. ;-)

But there should be a notice to your notation of your web-address.
In your emails you wrote: "http://home.att.net/~diannashipman",=20
this should be clickable but causes navigation errors sometimes.
This is, because the web path is incomplete. The complete address
I think is "http://home.att.net/~diannashipman/index.html", where=20
"index.html" is the default document and can be left and=20
"http://home.att.net/~diannashipman/" is the path.=20
If there is the last slash missing in the path some browsers,=20
sometimes can't find the address (once they find, once they
find not, I don't know why).

What's going on is actually a bit more complex. If the final / is
omitted from a directory URL, HTTP requires that the server reject
it. They do this in different ways. Here is what Dianna's server says
when I fetch that URL with my web tester:

: w3cat +H http://home.att.net/~diannashipman
HTTP/1.1 302 Moved Temporarily
Server: Netscape-Enterprise/3.6 SP3
Date: Wed, 17 Nov 1999 15:32:36 GMT
Location: http://135.145.253.40/~diannashipman/
Content-length: 0
Content-type: text/html
Connection: close

It then drops the connection. If the client program recognizes that
Location: line and tries again, then the user thinks that the URL
"worked". Actually, it failed but the software figured out what was
wrong. The fetch took longer than necessary, and produced excess Net
traffic, because two connections were required to get it right.

With some servers, the client actually gets a "404 NOT FOUND" error,
or any of several other errors. So figuring out that the final slash
is needed isn't trivial, and different releases of different browsers
are likely to fail to solve the problem for various servers.

The "/index.html" isn't actually all that good an idea. This is what
might be called a "standard default". That is, the original web
servers (both the CERN and NSC servers) looked for this file by
default, but it was configurable. The apache server has copied this,
and most commercial servers (except for Microsoft's, of course ;-) do
the same. Most webmasters don't bother changing it, but they can.
It's just one line in a config file. So index.html isn't guaranteed
to work. It's better to just end with the slash, and let the server
figure out if the directory has a default HTML file.

(If I add a 'R' redirect option to my tester, it gets her page, but
there is no clue from the server that it is called "index.html".)

BEGIN:VCARD
VERSION:2.1
N:Mettler;Klaus
FN:Mettler, Klaus (LFUST)
TEL;WORK;VOICE:0711/126-1944
EMAIL;PREF;INTERNET:Xxxxx.Xxxxxxx@XXXXX.XXX.XXX.XX
REV:19990409T104152Z
END:VCARD

Interesting. I wonder how standard this is. I've seen a number of
them, and I've been thinking of writing my own little program to
extract the data and add it to my own database. I wonder if the lines
are actually documented anywhere?

Previous message: Warm-ups (Norah Link)
Next message: Warm-ups (Pia Walker)
Previous in thread: computer question (Sue Wartell ext. 3387)
Next in thread: computer question (Alex Tweedly)
A Django site.