Re: [xsl] using xsl:message with UTF-8 characters

[Abel Braaksma <abel.online@xxxxxxxxx>]

Michael Kay wrote:
> I don't know how good Java is at getting the encoding right, for example
> whether it will use a different encoding if you use configuration options
> such as "cmd /u" identified by Abel. I'll do some experiments.
>   

Java will choose the default encoding of the underlying system, which 
is, in the case of Windows, the codepage set in International and 
Regional settings. This codepage is never compatible with IBM-437 (or 
CP437) used for the command window, which is age old (1981). When the 
Regional settings are set to US or some Western European country, the 
codepage will default to CP1252 (windows-1252) (which is, like I said, 
incompatible with the codepage for the console, giving the weird 
characters in the U+0127+ range).

It is very awkward that Microsoft never chose to upgrade the default 
codepage of the DOS console to be the same as Windows, but you can set 
your default settings in the registry or in some system *.cmd file (I 
forgot the name) (but then again, you can't set it to default to 
whatever is in your Regional Settings...)

> In Saxon, xsl:message by default uses a Java Writer, whereas "normal" result
> documents use a Java OutputStream. 

I'd like to argue in favor of defaulting to a particular encoding 
instead (i.e., UTF-8), because now it's like a lottery how the 
underlying system will determine what codepage it becomes (and build 
once run everywhere does not mean 'run everywhere and act equally' 
anymore, which I consider a pity). But such a discussion would be better 
suited on the Saxon list I believe.