"Kirill 'Big K' Katsnelson" <kkm(a)dtmx.com> writes:
Some time ago, Hrvoje Niksic wrote...
>You mean: binary-process-output is now ignored, and everyone should
>bind coding-system-for-read to `binary' instead. Don't you think this
>should be documented in NEWS, along with a rationale for the change?
Exactly. binary-process-output was NTEmacs influenced variable,
while coding systems are XEmacs-wode, not NT-only. The variable was
initially marked by marcpa as borrowed temporarily from NTEmacs,
while coding streams are not implemented. I do not really know if
removal of oblosete features should be put into NEWS[1], on your
discretion Hrvoje.
The purpose of NEWS is to document user-visible changes. If I have
code that does:
(let ((binary-process-output t))
...)
and that code no longer works in 21.2, this should well be mentioned
in NEWS, under "Lisp and internal changes". The entry might look like
this:
** The variable `binary-process-output' is now obsolete. This
variable was previously used to tell NT XEmacs to forgo the CRLF
detection and accept its input as binary. The correct way to do it is
now the same as under Unix, by binding `coding-system-for-read'.
The NT-specific code that looked like this:
(let ((binary-process-output t))
... code ...)
should be changed to this:
(let ((coding-system-for-read 'binary))
... code ...)