David Lichteblau <david@...> writes:
> ... a well-known bug, I believe. In SBCL, (subtypep 'sb-sys:fd-stream
> 'file-stream), even though not every fd-stream has a pathname. On Linux
I have sketch, and maybe 10% of the implementation for something that
gets rid of this:
* Replace FD-STREAMS with a vaguely java.nio-like low-level interface
for talking to the OS about bytes.
* Make FILE-STREAM encapsulate rather then subclass the above.
* Hack ANSI-STREAM to encapsulate an ANSI-STREAM-CLASS rather then
directly hang on to appropriate functions.
At this stage I'm fairly confident that I can finish this before the
year is out, but I have no idea about what the performance of this will
be, so holding breath is not recommended.
-- Nikodemus Schemer: "Buddha is small, clean, and serious."
Lispnik: "Buddha is big, has hairy armpits, and laughs."