[xslt-conformance] Issue statement regarding encoding

[David Marston/Cambridge/IBM <david_marston@us.ibm.com>]

A couple months ago, I was talking with the Microsoft guys (Pranav
and Kirill) about our respective submissions. We discovered that Xalan
creates UTF-8 encoded output by default, while MSXML creates UTF-16
encoded output by default. Neither team has been making it a habit to
include xsl:output directives to force the encoding. Thus, most of
our reference outputs are UTF-8, and most of theirs are UTF-16.

If I'm going to try their tests on Xalan for informal purposes, just
to evaluate what the test is doing, I'll probably insert an xsl:output
statement into each of their stylesheets that doesn't already have
one. In the context of this Committee, though, that would mean I was
altering their test.

Will our output comparison be able to compare UTF-8 and UTF-16
adequately? If not, how do we want to work around the defaults?

In a strict sense, we also have the issue that every test case that
lacks an xsl:output encoding declaration invokes one of these
http://www.oasis-open.org/committees/xslt/xsltdisc.htm#default-encoding-UTF-8
http://www.oasis-open.org/committees/xslt/xsltdisc.htm#default-encoding-UTF-16
discretionary items. We could flag every such test, but I think it's
preferable to discuss our real solution before embarking on that path.

This should be on the agenda for our next meeting.
.................David Marston