Robert Dodier writes:
> the one unresolved problem which i believe we should fix before
> releasing 5.9.2 is the placement of newlines.
[...]
> this topic seems to have been discussed most recently here:
> http://www.math.utexas.edu/pipermail/maxima/2005/thread.html#9091
>
> at present the problem is that there is a lack of newlines
> after stuff is printed by format, mtell, merror, or other means.
If I may summarize the changes which I proposed:
(1) Bind the standardized I/O customization variables to a single
synonym-stream (whose symbol is *top-io*) which may refer to a
socket-stream or to the value of *terminal-io* or to some other
stream.
(2) A simple-minded Gray-stream wrapper (tty-io.lisp) around
*terminal-io* addresses the newline problem in CMUCL and SBCL. This
change depends on (1).
I still think that (1) might be a good thing since, as I wrote
elsewhere in more detail, it fixes some issues (notably dribbling) and
unifies the overall setup (the original motivation for (1) was to
clean up CONTINUE from extraneous stuff in view of other changes). On
the other hand, the effect of this change is inherently
implementation-dependent, so some feedback from people actually
testing it would be welcome.
As for (2), it seems to work well enough, but I am still suspicious
because it seems a bit too simple-minded. Also, there's nothing it
can do for GCL. And it's additional stuff to maintain just for the
sake of some embellishment.
Personally, I find missing newlines much more annoying than blank
lines, so, frankly, if we are in a hurry for the release, I'd suggest
to revert Viktor's patch in the release branch.
Meanwhile, committing the stuff described here to the development
branch will be the best way to get some feedback, I guess, so I'm
preparing this now.
Wolfgang