[Mod_gzip] "mod_gzip_send_vary=Yes" disables caching on IE? (1.3.26.1a)

[Slava Bizyayev]

Hi Everyone again!

Yes Kevin, I'm listening. Just can't respond during the work hours...

Let me state clear once again: I'm experiencing local caching problems with
uncompressed content on M$IE-6.0 in presence of pure "Vary: *". The problem
disappears when I use "Vary: User-Agent,*" instead. However, semantically
these headers are identical (at least should be).

Rob is talking about HTTP headers in terms of proxy rules, They are not
implied to be the same for the end user's browser, which is associated with
one PC and one user only. In my understanding the caching mechanism of the
normal browser should not depend on Vary at all. In fact, it works a
different way sometimes (at least for M$IE-6.0 which I tested). Fortunately,
we have a way around in form of "Vary: User-Agent,*". It works fine for me
(I mean Apache::Dynagzip).

I agree with your main strategy, to go toward the requirements which Jordan
has from the position which I stay now. Unfortunately, I can't go ahead
right away: my former employer (Outlook Technologies, Inc.) has cut my
development server devl4 from the Internet today. The situation is out of my
control, I can't estimate the length of this outage. I will be back with you
as soon as I can...

Regards,
Slava


----- Original Message -----
From: <TOKILEY@aol.com>
To: <mod_gzip@lists.over.net>
Sent: Monday, December 09, 2002 2:50 PM
Subject: Re: [Mod_gzip] "mod_gzip_send_vary=Yes" disables caching on IE?
(1.3.26.1a)


>
> > That IE treats "Vary: Accept-Encoding" on (only) uncompressed responses
as
> > apparently meaning "don't cache". This wasn't an issue with older
versions
> > of mod_gzip because they did not send this header.
>
> Super. Then let's focus on just this one issue.
>
> If it turns out that MSIE is refusing to cache ANY kind of
> non-compressed response with any kind of "Vary:"
> header then so be it but for our purposes let's just focus
> on whether anything ( mod_gzip, mod_deflate, DynaZip, whatever )
> is going to be able to use "Vary: Accept-Encoding" with
> MSIE and non-compressed variants.
>
> I still think the fact that MSIE really MUST cache the
> 'compressed' responses because it MUST have the
> disk file to do the decompression is somehow
> confusing the issue but let's try and nail down
> exactly what MSIE 'standard behavior' is for
> non-compressed responses that DO have
> some kind of "Vary:" header ( Other than "Vary: *"
> which we know no one is really ever going to Cache ).
>
> >> (Jordan?) says
> >> that MSIE isn't caching ANY of the non-compressed variants
> >> if they have ANY kind of "Vary:" header.
> >
> >Yes, that is my experience. It always reloads the pages when navigating
the
> >history list (using the Back and Forward buttons), and it always reloads
> >them when you visit them later -- no If-Modified-Since request header is
> >sent (if the last visit had a Last-Modified response header), and any
> >Expires response header from the last visit (that was non-compressed and
> >contained a Vary header) isn't respected.
>
> Ok... clear enough.
>
> However... I believe Slava (?) has said he is NOT seeing
> this behavior. Let's ping him and verify that and if he's
> NOT seeing same problem let's figure out why the
> heck you are.
>
> >> What (Exact) versions and build levels of MSIE are
> >> we talking about here and does one really actually
> >> WORK while another does NOT?
> >
> >With this test:
> >
> >header("Vary: Accept-Encoding");
> >header("Expires: Mon, 08 Dec 2003 00:00:00 GMT");
> >echo "Hi";
> >
> >Here are the versions I tested. All of them "did not work" -- they reload
> >the page every time.
> >
> >- IE 6.0 + all latest updates:
> >  6.0.2800.1106 ;SP1;Q328970;Q324929;
> >- Plain IE 6.0:
> >  6.0.2600.0000
> >- IE 5.5 SP2:
> >  5.50.4807.2300 ;SP2;
> >- Plain IE 5.0:
> >  5.00.2920.0000
>
> That pretty much covers it.
>
> If anyone is using any version of MSIE any older than
> this list then they are pretty much crazy just because
> of all the security flaws alone in any version less than 5.0.
>
> Slava?... are you still reading the thread?
>
> What version(s) of MSIE are you using and are you seeing
> the same 'no cache' behavior?
>
> >> Where does Netscape fit into all this?
> >
> > I haven't personally tested this.
> >
> > Jordan Russell
>
> Neither have I. I don't use Netscape anymore and I
> don't know anyone who does.
>
> I have heard that the new Netscape 'Compact' version
> is much more stable than full version of Netscape but
> I have not tried that browser yet.
>
> Yours...
> Kevin
>
> _______________________________________________
> mod_gzip mailing list
> mod_gzip@lists.over.net
> http://lists.over.net/mailman/listinfo/mod_gzip
>