Am Mo., 20. Apr. 2020 um 18:34 Uhr schrieb Shiro Kawai <xxxxxx@gmail.com>:

[...]

> On Mon, Apr 20, 2020 at 6:27 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:

[...]

>> Don't we only have to save the input if the C reader sees a custom>> port? Buffering of the input read so far could even become part of the
>> port and not of the reader. The reader requests a handle and the port
>> starts to buffer under this handle. A non-custom port can simply
>> ignore those reader requests.

The more I think about it, the better I like the idea to delegate this
functionality to the port, at least insofar the implementation is
concerned. Every textual input port could have an internal "read"
method (in the OOP sense), every textual output port an internal
"write" method. The user-facing `read' and `write' procedures simply
delegate to these port methods. For custom ports, they have to do a
bit more work (like saving the characters read so far), for ordinary
ports, nothing changes.

This does not solve the problem with record printers written in C, but
such writers are implementation-specific anyway so their behavior is
implementation-dependent anyway.

-- Marc