Shibboleth Developers

["Adam Lantos" <>]

Scott,

On Thu, Oct 30, 2008 at 6:44 PM, Scott Cantor 
<>
 wrote:
> Alright, as usual when Java's involved, I'm wrong, get a headache, and end 
> up feeling dumber than when I started, so I'll just shut up.

It's okay, good to have such conversations sometimes.

> For the record, I don't necessarily think that the lower level libraries 
> should be required to rely on UTF-8. I do much the same thing, but always 
> considered that a flaw.
>
> Ideally, it's configurable, but mainly I think it's important that the 
> surrounding locale shouldn't influence the process if that's possible. If 
> the code knows it needs to handle data using a particular encoding, it 
> should probably be explicit about it and not rely on the Java locale to be 
> set correctly.

Sure, that was my instinct too... The xml marshalling code knows that
it produces utf-8 transfer encoded data, so it should do it explicitly
(and independently from the locale). Fortunately this modification
only involves the getBytes() methods before the base64 encoding
(AFAIK) in OpenSAML encoders, but honestly I do not see if this can
introduce problems somewhere else.


> The SP never relies on localized functions if it knows data has to be 
> handled as Unicode.
>
> -- Scott
>
>
>


cheers,
 Adam