Menu
▾
▴
#95 resetreduce clears too much
I tried to use the resetreduce command from the reset package. After that, trying to load a reduce file with the "in" comand gave the message:
Declare IN operator ? (Y or N)
also, the same happens with many other reduce internal commands, e.g. "operator", "df".
This is version 4859 on linux.
Are you using CSL REDUCE? If so, I suggest you try PSL REDUCE.
On Windows, I see this problem in CSL REDUCE but not in PSL REDUCE. I also don't see it in REDUCE on SBCL, so I think it's something specific to CSL.
Francis
Related
Bugs:
#95I am using CSL reduce. It would be quite convenient especially in the
GUI version to use resetreduce, to make reduce forget calculations,
but not the command line (e.g., load an edited version of the same
file). PSL reduce at the moment does not have command line editing, so
my use case is mostly pointless with that.
Also, I tried resetreduce with PSL reduce. In that case, it did not
clear variables I set (in a file loaded with the "in" command). So it
seems, resetreduce removes too much in CSL, and not enough in PSL.
On Wed, Feb 27, 2019 at 2:28 PM Francis Wright
fjwright@users.sourceforge.net wrote:
Related
Bugs:
#95I looked into this briefly. The "resetreduce" operation resets a load of
flags back to what it thinks should be their default values. These
defaults were selected quite a while ago, and one of them is that the
!*raise flag is reset to T, as if "on raise;" was performed. In CSL this
causes all input to be folded to UPPER CASE hence the word "IN" in capital
latters in the original bug report. If one then says "!o!f!f !r!a!i!s!e;"
it is possible to reset the sane state where lower case letters are left
alone.
In PSL the raise flag goes not cnvert things to upper case but instead it
converts to whatever case most internal symbols are in. Many years ago
that was upper case, these days it is lower case. CSL felt that that was
an odd use of a name, and so CSL has a flag called !lower to fold case
downwards.
I suspect but have not checked that for Common Lisp much of the interbal
world has to be spelt in UPPER CASE because at least early versions of
Common Lisp insisted that the name was eg CAR and was not car, and it
tended to case fold up on input and down on output in a way that
seruiously compromises any idea of case sensitive usage.
So I have put in a 1-line patch in packages/misc/reset.red that just
comments out the line that leads to !*raise getting reset to something
that is perhaps best described as historical.
As to the rest of the behaviour, those interested csn look into the source
file there and consider what they think should be reset and what preserved
and perhaps propose updates.
Arthur
On Wed, 27 Feb 2019, arpi wrote:
Related
Bugs:
#95You might be interested in "redfront", which provides bash-like history, search, and editing. For instance, call rfpsl.
Francis: Even prop() and quit are affected. The resetreduce code is probably way older than the CSL model with extensive compilation to C.
Last edit: Thomas Sturm 2019-02-27
Problem corrected.