Mailing List Archive

On Sat, 7 Oct 1995, Rob Hartill wrote:

> Everyone playing with the new Netscape beta?

I imagien not everyone...

> There's stuff there for language encoding and it
> sends "CONNECTION: Keep-Alive" too.

Yes. It also sends a Host: header. Someone should note to the http-wg
that it sends PQDNs and ports (apparently whatever's in between the first
and second /, it seems), but I'm too scared to do so (*smirk*).

At any rate, it seems to me it's time Apache thought about supporting these
things.

> ... it all looks very impressive (try view source)

--/ Alexei Kosut <akosut@nueva.pvt.k12.ca.us> /--------/ Lefler on IRC
----------------------------/ <http://www.nueva.pvt.k12.ca.us/~akosut/>
The viewpoints expressed above are entirely false, and in no way
represent Alexei Kosut nor any other person or entity. /--------------

> Yes. It also sends a Host: header. Someone should note to the http-wg
> that it sends PQDNs and ports (apparently whatever's in between the first
> and second /, it seems), but I'm too scared to do so (*smirk*).

Crikey! What ever happened to "get it right the first time"?
I wonder what happens when they access

http://www.apache.org

(no trailing slash) -- probably a core dump.

I just wish the Navigator crew would spend a little more time talking
(and listening) to the Netsite crew -- for some strange reason all
browser developers (except Bill Perry) read protocol specifications
with beer-bottle glasses.

.....Roy (or at least that's my polite theory of browser development)

>
>
> Everyone playing with the new Netscape beta?

I'd love to, but where is it?

>
> There's stuff there for language encoding and it
> sends "CONNECTION: Keep-Alive" too.
>
>
> ... it all looks very impressive (try view source)
>

--
Ben Laurie Phone: +44 (181) 994 6435
Freelance Consultant Fax: +44 (181) 994 6472
and Technical Director Email: ben@algroup.co.uk
A.L. Digital Ltd,
London, England.

Ben Laurie writes:

> > Everyone playing with the new Netscape beta?
>
> I'd love to, but where is it?

ftp[1-7].netscape.com/2.0beta

> > There's stuff there for language encoding and it
> > sends "CONNECTION: Keep-Alive" too.
> >
> > ... it all looks very impressive (try view source)

Definately, though java support is pretty weak. And for a litte fun

about:hype

most of you should recognize where thats from and who's speaking.

--
Ryan L. Watkins email: vamp@dnx.com
HTML Hacker / System Admin url: http://www.dnx.com/vamp/vamp.html
Dimension X, Inc. phone: 415.296.0100

>
> Ben Laurie writes:
>
> > > Everyone playing with the new Netscape beta?
> >
> > I'd love to, but where is it?
>
> ftp[1-7].netscape.com/2.0beta
>
> > > There's stuff there for language encoding and it
> > > sends "CONNECTION: Keep-Alive" too.
> > >
> > > ... it all looks very impressive (try view source)
>
> Definately, though java support is pretty weak. And for a litte fun
>
> about:hype
>
> most of you should recognize where thats from and who's speaking.
>
> --
> Ryan L. Watkins email: vamp@dnx.com

Thanks. I got it. But, sigh, it dumps core at startup (on Irix 5.3).
I guess I'll have to try the BSDI version. After that, I run out of crossover.

--
Ben Laurie Phone: +44 (181) 994 6435
Freelance Consultant Fax: +44 (181) 994 6472
and Technical Director Email: ben@algroup.co.uk
A.L. Digital Ltd,
London, England.

Got it, played with it... the mailer interface is cute once you figure
out the counterintuitive things about it. However, the java support
always makes it dump core when invoked, even after I've put the moz2_0.car
file *everywhere* that it's advertised to look. (In any case, a syscall
trace shows it dying on a segfault before it even tries to open or stat a
file in any of those locations, or to fork off a subprocess --- the only
two forks visible are for the 'ps' and 'netstat' which are part of the
new crypto key generation, and both those processes are unambiguously
dead by the time this gets going).

My last thought was that it might be X server incompatibility (we use
MIT X11, and they might have tested only against OpenWindows). However,
it doesn't work with OpenWindows either.

If anyone gets an applet to successfully display on the SunOS version,
no matter how trivial, I'd appreciate hearing about it...

(FWIW, the test cases I'm using are the example applets from the Sun
pre-beta SDK docs --- I generally use "hangman" as my example, but I've
tried other, simpler ones and they also bomb out).

rst

>
> Got it, played with it... the mailer interface is cute once you figure
> out the counterintuitive things about it. However, the java support
> always makes it dump core when invoked, even after I've put the moz2_0.car
> file *everywhere* that it's advertised to look. (In any case, a syscall
> trace shows it dying on a segfault before it even tries to open or stat a
> file in any of those locations, or to fork off a subprocess --- the only
> two forks visible are for the 'ps' and 'netstat' which are part of the
> new crypto key generation, and both those processes are unambiguously
> dead by the time this gets going).
>
> My last thought was that it might be X server incompatibility (we use
> MIT X11, and they might have tested only against OpenWindows). However,
> it doesn't work with OpenWindows either.
>
> If anyone gets an applet to successfully display on the SunOS version,
> no matter how trivial, I'd appreciate hearing about it...
>
> (FWIW, the test cases I'm using are the example applets from the Sun
> pre-beta SDK docs --- I generally use "hangman" as my example, but I've
> tried other, simpler ones and they also bomb out).
>
> rst

Well I've got it running on BSDI - does anyone know where I can pick up some
Java to test?

Cheers,

Ben.

--
Ben Laurie Phone: +44 (181) 994 6435
Freelance Consultant Fax: +44 (181) 994 6472
and Technical Director Email: ben@algroup.co.uk
A.L. Digital Ltd,
London, England.

Go to the Java home page (http://java.sun.com), then go to the docs on the
pre-beta SDK release --- that has pointers to a page of sample applets. NB
these are the *only* applets on the site which are supposed to work with
Netscape; most current applets are designed to work with last spring's alpha
release of the Java technology, and there have been incompatible changes
since.

As I said, my test case is generally hangman, but I've tried others, and I
can't get any of 'em to work.

Off to lunch...

rst

Ooops --- one other note. My distribution came with a file, moz2_0.car,
which must be placed somewhere that Netscape can find it, or it is
*advertised* to core dump when it tries to run an applet. I presume the
BSDI release works the same way; check the README for what to do about
this. (If you're running Netscape from a directory which has a copy of
the thing, "setenv CLASSPATH ." before running netscape is supposed to
be enough --- again, at least on SunOS).

That's all that's supposed to be necessary to make it work, but since
it hasn't worked for me, either the directions are missing something, or
I am.

rst

Ben Laurie writes:

> Well I've got it running on BSDI - does anyone know where I can pick up some
> Java to test?

You cant, cuz the BSDI version doesnt support java yet. Release notes
indicate its only (supposedly) supported on Solaris, Windows95/NT,
Iris and SunOS. Havent managed to get it to run java under a Solaris
2.5b machine (either the 2.3 or 2.4 version), but it does run under
2.4 and Win95. Windows version is dog slow compared to HotJava.

For testing I'd try http://www.javasoft.com/JDK-prebeta1/applets/index.html.
I tend to try the wireframe demos.

Of course this nolonger has anything to to with apache ...

--
Ryan L. Watkins email: vamp@dnx.com
HTML Hacker / System Admin url: http://www.dnx.com/vamp/vamp.html
Dimension X, Inc. phone: 415.296.0100

AS someone said, it doesn't do Java on BSDI. I guess I'll have to wait for
someone to fix the Irix version. Or get the Win95 version (but I didn't see
that on the mirror I used). One thing, though - the cacheing problem I was
going on about a while back (not working with client pull) seems to be fixed!

Cheers,

Ben.

--
Ben Laurie Phone: +44 (181) 994 6435
Freelance Consultant Fax: +44 (181) 994 6472
and Technical Director Email: ben@algroup.co.uk
A.L. Digital Ltd,
London, England.

For anyone with an interest in my travails, I finally got the Java support
to work, more or less.

It turns out that the Sun Java support in Netscape does not work at all on
sun4c machines (running either SunOS or Solaris --- I finally got this hint
by reading an article on netnews reporting this for Solaris, and quickly
confirmed it for SunOS). I tried it out on a Sparc 5, though, and on that
machine it almost works (the hangman applet still doesn't work, but it
doesn't crash netscape, and most of the other examples I've tried do seem
to more or less be there).

Time to try out that bug report form, I guess...

rst

> (FWIW, the test cases I'm using are the example applets from the Sun
> pre-beta SDK docs --- I generally use "hangman" as my example, but I've
> tried other, simpler ones and they also bomb out).

Same problems. Solaris 2.4 version coredumps before anything happens with
any Java applet.

-Max

On Sun, 8 Oct 1995, Max Levchin wrote:

> > (FWIW, the test cases I'm using are the example applets from the Sun
> > pre-beta SDK docs --- I generally use "hangman" as my example, but I've
> > tried other, simpler ones and they also bomb out).
>
> Same problems. Solaris 2.4 version coredumps before anything happens with
> any Java applet.

Can someone tell me what the Java implementation (or mis-implementation)
in Netscape Navigator 2.0b1 has to do with the Apache HTTP Server please?
Aren't there other forums for this sort of thing? Not that I really mind,
but...

Now, as I see it, there are several things to note about Netscape 2.0b1
that are relevent to our work here, and I will note them. I have access
only to the Mac version, so everything I say is relavant to that
platform's Netscape Navigator, but I imagine it's applicable to all
three:

* It sends Connection: Keep-alive. However, I have no clue if it supports
it. Is there any way to test this? At any rate, this seems to have become
a de facto standard as of late, so maybe we should be supporting it.

* It sends a Host: header with every transaction. This is a good thing.
As I understand it, future versions of NCSA Mosaic will also send such a
thing. As I further understand it, future versions of NCSA httpd will
support receiving such a thing. I would submit Apache should be supporting
such a thing as well. Like virtual hosts. But different.

* One positive note with regards to content negotation: It seems no longer
sends Accept: */* when requesting inline images. Just the four image types
(all one one line, too). It sends the */* when requesting real documents,
though. So this is good.

Hmm. Probably some other things, but I can't think of them.

--/ Alexei Kosut <akosut@nueva.pvt.k12.ca.us> /--------/ Lefler on IRC
----------------------------/ <http://www.nueva.pvt.k12.ca.us/~akosut/>
The viewpoints expressed above are entirely false, and in no way
represent Alexei Kosut nor any other person or entity. /--------------

> Can someone tell me what the Java implementation (or mis-implementation)
> in Netscape Navigator 2.0b1 has to do with the Apache HTTP Server please?

er nope.

> I have access only to the Mac version, so everything I say is
> relavant to that platform's Netscape Navigator, but I imagine
> it's applicable to all three:

Do the Mac/Windoze versions recognise
"Content-encoding: x-gzip" or .."x-compress" ?

Apache can send these, so I think the question is relevant to this list :-)


rob

On Sun, 8 Oct 1995, Rob Hartill wrote:

> Do the Mac/Windoze versions recognise
> "Content-encoding: x-gzip" or .."x-compress" ?

I can't speak about the Windows version, but Netscape 2.0b1 for Macintosh
does "recognise" it... sorta. A .html.gz file I just tried resulted in a
dialog box with:

Warning: unrecognized encoding 'x-gzip'.

Then it displayed garbage (which I'm assuming is the compressed file) in
the window. So it doesn't *decode* it. But it knows it's there, and is
smart enough to tell the user.

x-compress results in the same thing, with x-gzip replaced with x-compress.

> Apache can send these, so I think the question is relevant to this list :-)

Okay...

--/ Alexei Kosut <akosut@nueva.pvt.k12.ca.us> /--------/ Lefler on IRC
----------------------------/ <http://www.nueva.pvt.k12.ca.us/~akosut/>
The viewpoints expressed above are entirely false, and in no way
represent Alexei Kosut nor any other person or entity. /--------------

On Sun, 8 Oct 1995, Alexei Kosut wrote:
> On Sun, 8 Oct 1995, Rob Hartill wrote:
>
> > Do the Mac/Windoze versions recognise
> > "Content-encoding: x-gzip" or .."x-compress" ?
>
> I can't speak about the Windows version, but Netscape 2.0b1 for Macintosh
> does "recognise" it... sorta. A .html.gz file I just tried resulted in a
> dialog box with:
>
> Warning: unrecognized encoding 'x-gzip'.
>
> Then it displayed garbage (which I'm assuming is the compressed file) in
> the window. So it doesn't *decode* it. But it knows it's there, and is
> smart enough to tell the user.
>
> x-compress results in the same thing, with x-gzip replaced with x-compress.

Shit - I thought they were to have gunzip and uncompress in 2.0 mac and
windows.

More relevance - right now since none of the browsers that can accept
gzip'd files tell the server via Accept-Encoding: headers, it's
impossible for a server to "do the right thing" with a .gz file (i.e.
gunzip it for non-gz-unzipping clients server-side) except leave it up to
the user/interface ("click *here* for the gzipp'd version"). This is
crucial in VRML for example.

Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/

>
> On Sun, 8 Oct 1995, Max Levchin wrote:
<snip>
> * It sends Connection: Keep-alive. However, I have no clue if it supports
> it. Is there any way to test this? At any rate, this seems to have become
> a de facto standard as of late, so maybe we should be supporting it.

I believe the Navigator supports the same protocol we (NCSA, Mosaic & HTTPd)
and Spyglass do. We've been pushing this as more server friendly than the
multiple connection approach that is also very popular.

> * It sends a Host: header with every transaction. This is a good thing.
> As I understand it, future versions of NCSA Mosaic will also send such a
> thing. As I further understand it, future versions of NCSA httpd will
> support receiving such a thing. I would submit Apache should be supporting
> such a thing as well. Like virtual hosts. But different.
<snip>
> --/ Alexei Kosut <akosut@nueva.pvt.k12.ca.us> /--------/ Lefler on IRC

Brandon managed to get the host support in the current beta (1.5b6).

-Beth

--
Elizabeth(Beth) Frank
NCSA Server Development Team
efrank@ncsa.uiuc.edu

> > Yes. It also sends a Host: header. Someone should note to the http-wg
> > that it sends PQDNs and ports (apparently whatever's in between the first
> > and second /, it seems), but I'm too scared to do so (*smirk*).
>
> Crikey! What ever happened to "get it right the first time"?
> I wonder what happens when they access
>
> http://www.apache.org
>
> (no trailing slash) -- probably a core dump.
>
> I just wish the Navigator crew would spend a little more time talking
> (and listening) to the Netsite crew -- for some strange reason all
> browser developers (except Bill Perry) read protocol specifications
> with beer-bottle glasses.
>
> .....Roy (or at least that's my polite theory of browser development)

Heh, to be fair the NS2b release was a panic/coredump/traceback/lawsuit
reaction to the security fukup. The bug list is probably the single most
impressive part of the release. It's certainly not a mature product. Still,
if Apache Group had released code like that noone would bat an eyelid.

...if

Ay.

On Sat, 7 Oct 1995, Roy T. Fielding wrote:

> Crikey! What ever happened to "get it right the first time"?
> I wonder what happens when they access
>
> http://www.apache.org
>
> (no trailing slash) -- probably a core dump.

Running a web "server" (really just a perl script that listens on port 80
and displays what it gets) on www.dal.net, and requesting the following
URLs with Netscape 2.0b1 for Macintosh, I get the following results:

http://www.dal.net/
Host: www.dal.net

http://www.dal.net
Host: www.dal.net

http://www/
Host: www

http://www
Host: www

http://www.dal.net:80/
Host: www.dal.net:80

http://www.dal.net:80
Host: www.dal.net:80

http://www:80/
Host: www:80

http://www:80
Host: www:80

Perhaps someone should kindly note to Netscape that this is not the
perferred behavior. I think. Although, I must note, up until a couple days
ago, http-wg still appeared to be debating the very same issues, so
perhaps Netscape will have changed it for 2.0b2 already?

> I just wish the Navigator crew would spend a little more time talking
> (and listening) to the Netsite crew -- for some strange reason all
> browser developers (except Bill Perry) read protocol specifications
> with beer-bottle glasses.

> .....Roy (or at least that's my polite theory of browser development)

--/ Alexei Kosut <akosut@nueva.pvt.k12.ca.us> /--------/ Lefler on IRC
----------------------------/ <http://www.nueva.pvt.k12.ca.us/~akosut/>
The viewpoints expressed above are entirely false, and in no way
represent Alexei Kosut nor any other person or entity. /--------------