📝Add notes for transcode_view steps later

documentation/source/future.rst

@@ -35,13 +35,6 @@ This is where the status and progress of the library will be kept up to date. Yo
 
 
 
-Document Conversion
--------------------
-
-Conversion routines need to be documented, alongside their extension points. They're not yet documented because it's currently a toss up between plain ol' ADL vs. tag-invoke. We'll probably invent a tag-invoke based solution, to prevent random functions from catching on. This also means we wouldn't need to prefix all extension points with C-style naming (currently, it's ``text_foo`` or ``ztd_text_foo``).
-
-
-
 Copyable State
 --------------
 
@@ -51,6 +44,18 @@ Right now, all state parameters are assumed to be move-only. This is detrimental
 
 
 
+Transcoding Iterators/Transcode View
+------------------------------------
+
+Right now these types would not work especially well for input and output ranges. They should be modified just like the internal ``ztd::text::__txt_detail::__encoding_iterator`` class types, so that they work with ``input_iterator`` and ``output_iterator`` types.
+
+- ☐ Improve constructor delegation and make sure to explicitly implement default construction vs. letting it happen with ``=default`` (which does not work for some of the base types present).
+- ☐ Modify implementation to cache data and position when an input or output iterator is detected.
+- ☐ Return ``const value_type&`` for ``reference`` to enable C++20 ranges to work properly.
+- ☐ Mark as ``enable_borrowed_range`` when C++20 is detected.
+
+
+
 Normalization
 -------------