[Already reported Perl Bug] Confusion over utf8 and in-memory filehandles

UPDATE (solved fsvo)

It turns out this a known bug, which surprisingly has not been fixed yet. In the meantime "do not do this if it hurts" seems to be the best course of action :(

/UPDATE

Greetings venerable monks,

While I think (humbly) I have a rather good grasp of how unicode is handled by perl in and out, I find myself stumped by the following example:

perl -e '
  my $str = { map { $_ => "\x{A9}" } qw(byte char) };
  utf8::upgrade($str->{char});
  for (keys %$str) {
    open (my $fh, "<", \do{$str->{$_}});
    printf( "$_ is read as %s\n", unpack "H*", <$fh>);
  }
  printf "Strings are: %s\n",
    ($str->{byte} eq $str->{char} ? "equal" : "different")
  ;
'
[download]

I understand why "char" and "byte" are considered equal. What I do not understand is why the internal storage details "leak" through the in-memory filehandle.

Explanations welcome!

Comment on [Already reported Perl Bug] Confusion over utf8 and in-memory filehandles Download Code
Re: Confusion over utf8 and in-memory filehandles by Anonymous Monk on Dec 28, 2012 at 00:56 UTC
What do you mean, how do they leak through? `byte is read as a9 char is read as c2a9 Strings are: equal` [download] \xa0 is U+00A9 is ord 169, c2a9 is ord 169 utf8-encoded, after you decode it, it is chr 169	[reply] [d/l]
Re^2: Confusion over utf8 and in-memory filehandles by ribasushi (Monk) on Dec 28, 2012 at 06:46 UTC
It is the same string with the utf8 flag flipped up. As such I would expect the same bytes to be available when reading it as a "filehandle". Yet the utf8-ness (which is claimed to be an internal impl. detail all ovetr the docs) is "visible" in this case. I am not sure I understand why (nor can find any relevant perldoc)	[reply]
Re^3: Confusion over utf8 and in-memory filehandles by Anonymous Monk on Dec 28, 2012 at 07:01 UTC
Oh, I get it now. Some discussion of it in #109828: PerlIO::scalar does not handle UTF-8	[reply]
Re^4: Confusion over utf8 and in-memory filehandles by ribasushi (Monk) on Dec 28, 2012 at 07:45 UTC


"be consistent"
	PerlMonks