W liście z pon, 16-08-2004, godz. 17:00 +0300, Jarkko Hietaniemi
napisał:
> > I mean XS. If SvUTF8 is false, I don't know whether to interpret the
> > contents as ISO-8859-1 or according to the locale.
>
> True. But if you know nothing of where the SVs are coming you would not
> know it anyway, I think. You cannot not know whether the bytes in the
> SV are characters at all, but instead a binary pack() buffer or a vec()
> bitvector, for example.
Because of this I don't convert Perl scalars to Kogut strings
automatically by default. They can be explicitly converted to byte
arrays or numbers etc. or just wrapped as a whole without converting
their contents.
But when the user of the bridge, who wants to use some Perl package
from Kogut, wants to convert a Perl scalar to a Kogut string and told it
explicitly, I must determine actual Unicode code points, because Kogut
strings are Unicoded.
I can assume that a non-UTF-8 scalar is encoded in ISO-8859-1, which -
as I conclude now - was the intended meaning. Unfortunately too often it
will actually be not the case. Maybe with the right set of pragmas Perl
code from various packages will actually consistently work that way.
Do these pragmas influence the behavior of code imported from packages?
--
__("< Marcin Kowalczyk
\__/ [EMAIL PROTECTED]
^^ http://qrnik.knm.org.pl/~qrczak/
