Open DarkArc opened 10 months ago
I agree that the partition const.str
has morphed over time to serve more generally as backing store for all character strings with the encoding moving into the entry of the character string descriptor. It makes sense to reconsider that and introduce dedicated partitions:
const.str.utf-8
const.str.utf-16
const.str.utf-32
That would probably means replacing the uses of TextOffset
for string literals with an StrIndex
with sorts:
StrSort::Utf8
StrSort::Utf16
StrSort::Utf16
TextOffset would be an index into const.str.utf-8
since all names are internalized as UTF-8 identifiers.
It's currently quite difficult to implement visualization for a given
const.str
entry as the partition contains several possible character encodings and does not specify what character encoding is used for a particular element of the partition (instead relying upon the context for how the string is referenced (e.g.,src.word
can reference aconst.str
entry as a wide character literal whileTextOffset
contextually means a UTF-8 null-terminated identifier).I think it would be a significant improvement if there was either a partition per-encoding (e.g.,
const.utf-8-str
,const.utf-16-str
, etc, or an encoding specified per line). This similarly would make it harder for folks -- such as myself :) -- to misinterpret the encoding of the contents of the partition.