Antwort: RE: Antwort: RE: [Mod_gzip] Empty output with all .php pages

[mod_gzip@lists.over.net]

Hi Matt,


>> Don't trust _any_ browser - look at the raw HTTP output,
>> using something like
>>      http://www.schroepl.net/cgi-bin/http_trace.pl
>>
> My apologies, it was the browsers (both IE and mozilla)
> I tested with which were causing the output to be munged.

They don't actually "munge" the output - they just don't
understand it (in your case), still try to parse it,
create some basic document tree, and display that docu-
ment tree (rsp. its HTML equivalent) when you invoke the
"view source" function.

Now that you had a glance at the data one step earlier,
you know where the problem lies, and can use the browser's
display function for what it's worth (and it is worth a
lot with Mozilla, as long as the browser actually found
something worth parsing).

> Thanks for the URL. I'm still stuck, therefore, as to
> what the problem I have that is causing the "DECLINED:HAS_CE".

Use "http_trace.pl" to inspect the HTTP headers of your
PHP page's output.
This will tell you which Content-Encoding is already
applied there ... it isn't reasonable to guess what's
going on when you have the option to _know_.

Maybe some configuration of your PHP installation makes
the page already be compressed? We still don't know
whether you even have any _problem_ at all.
You may still just misinterpret "HAS_CE" as being some
"error" (which it is not, actually - all it says is
"I don't dare to compress this content because someone
encoded it before and I'm afraid the browsers won't be
able to read it then").

By the way, has any user tried to _disable_ this check
(i. e. remove line 5814-5828 from "mod_gzip.c") and made
any tests with up-to-date browsers?
Maybe in the year 2003 there finally might be at least
_one_ browser that actually _understands_ HTTP/1.1, and
would allow more than one Content-Encoding applied? (My
best candidates for that being Mozilla 1.4 and Opera 7.1.)
Which would then call for another configuration directive
to only conditionally activate this check ...

Regards, Michael