Re: Using Perl's cc

[Eli Zaretskii]

> Date: Sun, 5 Jul 2015 15:50:01 +0100
> From: Gavin Smith <address@hidden>
> Cc: address@hidden
> 
> Do you think the choice of which functions should be used can be made
> at run-time, possibly in addition to a choice at build-time? For
> example, if switching to a UTF-8 locale fails?
> 
> There's two big hurdles to running the extension module: one is
> building it, which it looks like you will manage to do; but the other
> is loading it from a running Perl instance, and I still don't know if
> that will succeed. I did eventually get both to work on the OpenCSW
> test machines for Solaris 10
> (https://buildfarm.opencsw.org/buildbot/waterfall?category=texinfo),
> so it is promising that it worked on a system other than my own, but I
> would have to see it work on many more systems before I was confident
> it was reliable, especially on a system like Windows.

There's one other issue with the MS-Windows support: it's wchar_t is a
UTF-16 encoding of Unicode codepoints.  So characters outside of the
BMP yield 2 wchar_t values, not 1.  The code in xspara.c assumes
wchar_t is a scalar type that is wide enough to support any Unicode
codepoint, which is false for Windows.

To resolve this, we could use a data type other than wchar_t, and
convert the UTF-16 encoding into a Unicode codepoint with a trivial
function.  Is it OK to make that change?