EDS content negotiation in Chromium

Rod Widdowson rdw at steadingsoftware.com
Fri Mar 1 04:45:53 EST 2013


I'll brush off a non English VM (it will be windows) and try to reproduce
this with Chrome.

As I recall (and this was a painful part of the EDS) this is a bit of a
disaster area.   You see those nice setting that the browser gives you, well
they are for the server and there is no way to get hold of them in
javascript save getting the server to bounce them back down to you...

Various browsers provide more or less non-standard mechanisms to get a
language, but it tends to be the base language of the session the browser is
running in, not the language that the users specified in the browser.

So if this turns out to be reproducible it may not be fixable, or it may be
a long process  - or Google may show us the way...

Rod

> -----Original Message-----
> From: dev-bounces at shibboleth.net [mailto:dev-bounces at shibboleth.net]
> On Behalf Of Peter Schober
> Sent: 01 March 2013 09:05
> To: dev at shibboleth.net
> Subject: EDS content negotiation in Chromium
> 
> I find that using the EDS with Chromium (Chrome on GNU/Linux, X11) the
> language in the EDS does not change with what the browser offers as
> acceptable. (Granted there's currently only one additional language
> part of the distributed code.)
> 
> With a very old Chrome (v6, from Debian 6.0) I always get English,
> with a newer one (v18) I always get German, no matter the order in my
> browser. EDS code is current and the same in all cases.
> 
> Is someone able to reproduce that? I'll file an issue then.
> 
> Chrome v18 (According to http://wengeance.com/headerdump )
> Accept-Language: en,de;q=0.8
> or
> Accept-Language: de,en;q=0.8
> 
> Both also include the following, no idea of that matters:
> Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.3
> Accept-Encoding: gzip,deflate,sdch
> 
> Firefox v16, where EDS changes language as intended:
> Accept-Language: en,de;q=0.5
> Accept-Language: de,en;q=0.5
> 
> Accept-Encoding: gzip, deflate
> But no Accept-Charset, it seems.
> 
> -peter
> --
> To unsubscribe from this list send an email to dev-
> unsubscribe at shibboleth.net



More information about the dev mailing list