Mailing List Archive: pragma no-cache

pragma no-cache

Aug 1, 1995, 9:09 PM

Post #1 of 4 (289 views)

I think this might fix it....

http_protocol.c

if (status == USE_LOCAL_COPY)
return;

should be

if (status == USE_LOCAL_COPY) {
fprintf (c->client, "\015\012");
return;
}

We appear not to be adding the end of header \r\n for a 304 response...
Only the CERN proxy seems to mind, but it should be fixed. Sure beats
the Pragma nonsense.

rob

Re: pragma no-cache [ In reply to ]

hartill at ooo

Aug 1, 1995, 9:17 PM

Post #2 of 4 (295 views)

Permalink

>
> I think this might fix it....

Yup, tested at Cardiff via arachnophobia.mit.edu

Re: pragma no-cache [ In reply to ]

rst at ai

Aug 1, 1995, 11:20 PM

Post #3 of 4 (286 views)

Permalink

Rob --- that's what I'm seeing as well. The weird thing is that some
machines (web.mit.edu) fall down flat, while others (hensa) are fine.
I agree that this is a better fix --- the "reload-doesn't-reload"
behavior, which this leaves us with, is tolerable, given that it's
really Netscape's problem anyway.

I'm running code with pretty much your exact fix, and I've verified
that it fixes the problem (I've also diked out the Pragma: business).

I'm too zonked to trust myself to spin a release now; there will be a
new one early tomorrow, with this in it (I trust no one objects?).

rst

PS --- the system-dependancy of the proxy behavior is weird. It has
the flavor of a null-pointer dereference, which will kill it
on some machines but "work" on others...

Re: pragma no-cache [ In reply to ]

rst at ai

Aug 1, 1995, 11:24 PM

Post #4 of 4 (286 views)

Permalink

Proof that I'm zorched --- I got hensa's real web server confused with
their cache, which was why I thought they were both CERN. Clearly CERN
misbehavior (though under the circumstances I'm not sure that it's much
of a bug --- there will always be things that trip it up).

Clearly, I'm in no condition to release software...

rst