Jeremy, thanks for the info!
I see the section of http://ds.internic.net/rfc/rfc1738.txt you're
to, on page 7. I tried an experiment: with both
Netscape 3.x and Netscape 4.01 I fetched three different URL's:
I'll be danged if all three URL's didn't all return different results!
There was no difference between Netscape 3 and 4.
The first one errors out with "Netscape can't retrieve file or directory
the second one prints out a naked directory listing (no welcome
the third one prints out a decorated directory listing (with welcome
On Internet Explorer, it always adds the / to the URL, so #1 is the
same as #2;
and #2 and #3 both return a decorated directory listing (with welcome
I agree with Jeremy: Netscape probably needs to clean up their act.
Don't know if the workaround he suggests for ftp.unicode.org will work,
they should try it, considering as how Netscape is unlikely to fix this
This concludes this emergency test of the Extremely Pedantic Kibitzer
> From: Stairways Support[SMTP:email@example.com]
> Sent: Monday, July 28, 1997 12:08 AM
> To: firstname.lastname@example.org; email@example.com;
> Subject: Re: Bug in NetPresenz: can't fetch ftp://ftp.unicode.org
> withNetscap e
> >Probably the people who wrote NetPresenz haven't tested fetching
> >using ftp:/ URL's with Netscape, since NetPresenz is also an HTTP
> >but IMHO it's a bug worth fixing.
> >Dan Kegel
> Hi, Jeremy here, the Stairways support droid.
> NetPresenz is primarily an FTP server, and we've tested it. :) If it
> the bug I think it is, then it is a bug in Netscape's implementation
> of FTP.
> When an FTP client connects to an FTP site, it is logged into a
> which is not necessarily the root (top) of the directory structure-
> instance a user might be logged into his own directory, but still be
> to navigate up and back down into someone else area (with, for
> read only privileges).
> A single / indicates that the path follows from wherever the FTP
> gets dumped when it starts. A double / (//) indicates that you should
> the path from the very top of the accesible structure, the root.
> Anarchie and Fetch understand this and treat the URL correctly.
> (presumably) assumes that the root is always the directory you connect
> initially, so when it parses the URL it looks for the last // and
> follows is considered to be the machine.
> This is wrong: it should read from the first // after the :, not the
> //. (The :// pattern isn't guaranteed: protocol information can be
> between the : and the //.) So when netscape sees:
> ..it reads this as:
> ...and assumes the 'path' is a machine. Which is just plain wrong.
> You can get around this problem in NetPresenz by making sure that the
> directory people initially connect to is always the root, but
> basically it
> is a bug in Netscape.
> Jeremy, Support Droid.
> Stairways Software Pty Ltd <firstname.lastname@example.org>
> Web:<http://www.stairways.com/> FTP:<ftp://ftp.stairways.com/>
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:36 EDT