Shift-JIS interpreter does not handle U+3000 ideographic spaces correctly

[YunataSavior]

Example function from Twilight Princess in the dbg rom:

daNpc_Ks_HIO_c::genMessage

Example string:

"　小猿　"

[encounter]

What are the bytes and what should it decode to?

[LagoLunatic]

The bytes are:

	.4byte 0x81408FAC
	.4byte 0x898E8140

objdiff displays it like this, and also copies it to the clipboard like this, with the unicode sequence of the ideographic space instead of the space itself:
"\u{3000}小猿\u{3000}"
It should decode to:
"　小猿　"
Wonder if this is an encoding_rs bug.

[LagoLunatic]

I tried changing the symbol to be type data:sjis in symbols.txt and dtk correctly outputs the space in the asm:

	# 　小猿　
	.byte 0x81, 0x40, 0x8F, 0xAC, 0x89, 0x8E, 0x81, 0x40, 0x00

Since dtk also uses encoding_rs, it probably isn't an encoding_rs bug and is probably a bug in how I used it.

[LagoLunatic]

It seems the reason for this is that debug string formatting (:?) is used instead of regular string formatting, which causes characters that could be confused visually to be escaped. It doesn't seem to be limited to just Shift JIS, it already did the same with ASCII characters before Shift JIS support was added which is where I copied the debug formatting from.

So for example a string with a tab in it displays and copies as "\t" instead of " ". Is this intended behavior? Or do we want characters to be rendered as they are even if they're visually similar?