httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@liege.ICS.UCI.EDU>
Subject Re: type-map spin
Date Mon, 09 Dec 1996 01:24:35 GMT
> On Sun, 8 Dec 1996, Roy T. Fielding wrote:
> 
>> Ummm, what the heck is Content-Language doing with a qs value?
>> That doesn't excuse a crash, so I'd say a debug of the token-reading
>> code is in order.
> 
> I don't know *why* I did it.. After the server had not responded I went 
> to change it and my machine was getting slower and slower as swap grew... 

The question is: does removing it fix the problem?  I am at home right
now, so can't test it myself.

>> You might also try using qs=0.55 instead of qs=.55, etc.
> 
> What difference would that make?

.55 is not a legal q-value in HTTP, so our code might not have been tested
with .55 but definitely has been tested with 0.55.  I am just trying
to determine the problem case for testing.

> Also, Alexei said that the 'best' if none match is smallest... This is 
> not good in the case I am trying to do where gif should be sent if the 
> browser doesn't say it can handle jpeg... and jpeg is definately smaller :-)

Wouldn't the GIF match in that case?  I think you mean the best-if-all-equal
case, which is a problem with some browsers.  Basically, it requires a
separate default algorithm based on a BrowserMatch variable, but nobody
had the time to do it for 1.2.

.....Roy

Mime
View raw message