TEST
	Test 0
		Regression prevention: absolute -f canonicalization

	Test 1
		Kerning

		Mixtures of -m_odes and orientation (-[xX]), default mode

		Narrow & wide wrap

		overlap
			single column glyph e.g; term (esp. in Ransom)

			hardblanks

	Test 2
		-U=0, -U=-1

	Test 3
		Ransom -m_odes

	Test ?
		DOS-formatted txt file font

		bifig.flf post-processing, problem with (missing) hardspaces?

FEATURES
        More smushing modes, consult http://bolknote.ru/files/figlet/?

	"Vertical Fitting", indicated by header[-2] above

	Ransom support for ANSI & Unicode (w/ sparse codepoints)

	Ransom script to generate font, test output, then save font

        QOTW19 "OCR"? http://perl.plover.com/qotw/e/019

	controls; h j b g
		isn't u enough?

	other scripts: chkfont... this is hard

FIX
	Control
		literal needs to use $RE{UTFchar}
			$opts{-U} = 1 if /^\s*/
			$opts{-U} ? UTFchar : bytechar

		if pre 5.6 ignore char > 0xFF and use \x.. syntax?!
			switch to s alternation (for > 0xff)?

	XXX	cannot support unicode literals...?
			at all (#$#@, could translate to \x{} notation)

			just pre 5.6 (could translate to s///)
	Ransom
		Unicode support

	Native UTF-8 dependent (major rev bump and pre 5.8 support dropped)
	??	TOIlet permits widechars as EOL!

		charmapl.pl
			TOIlet fonts?

	$\ should be used for output, not $/?
	...although there may have been some emthod to this madness?
	This would be a major-ish change, and would wait for major rev bump

REFACTOR
	Switch to ::AnyInflate?

	Benchmark
		$font alias again

        Negative Char performance
		 Pre-allocating length halves size of array,
		 req. scan file for - 1st, seek/read backwards? iff -U
		     (assumes sane file order, but is okay for optimization?)
	XXX	     Alas header[-1], codetag count, is useless here

		 Hash(collision performance hit)?

		 2nd array(code complexity?)?

	detect if glyph has been seen before, and reference it?

	reduce reliance upon $/
		ammend split expression?

        Native UTF8 in 5.8+? Major version bump and drop support for $]<= 5.7

	-m=-0
		a) keep current hack on hack
		b) use header width

		may break, depending on Text::Wrap?

		use header[3] as initial approximation of char width,
		to help nix up-to 0x7e heuristic

	header	http://www.jave.de/figlet/figfont.html#headerline
		[-2] = full layout
	XXX	[-1] = figchar count!! optimize file parser with this?


	Revisit local 1.07 branches, anything to salvage?

	Experimental parsing (mis-)feature?
	my $file = qr/^(??{$hunk})*$/x;