If you look at the actual output file (whether it's XML or HTML, you
don't say) I'm sure you'll find it's byte-for-byte identical on both
systems. What happens after that is outside Saxon's control. You don't
say how you are viewing or printing the file, but you can be pretty sure
that's where the problem lies. There are two things to look for: either
the software that's doing the final display/printing is misinformed
about the encoding of the characters it is displaying, or the font it is
using for the display doesn't contain suitable glyphs.
The encoding of the source document and stylesheet are probably
irrelevant; the encoding of the result document may be worth looking at.
On 14/12/2010 23:05, RB45F wrote:
> I have a strange bug related to the use of a bullet character (U+8226). My
> stylesheet writes this character 3 different ways: by unicode character
> (i.e. pasted directly into xslt), by character entity ref, and by the
> codepoints-to-string((8226)) function. When saxon is run on Ubuntu linux all
> 3 work fine as expected.
> However when saxon is run on CentOS linux (which is used by Amazon EC2
> servers), the saxon xslt output from all 3 methods is "?????????" or
> sometimes just "???". Both my source document and stylesheet use xml
> declarations that specify UTF-8 encoding.
> Does anyone know what causes the output of "?????????" in an xslt result
> document? Perhaps knowing the reason this output is generated will provide
> me a clue as to what is happening.
> Many thanks for any help.