Re: Using history builtin with -ps options give unexpected results
From:
Martin D Kealey
Subject:
Re: Using history builtin with -ps options give unexpected results
Date:
Wed, 13 Aug 2025 21:34:11 +1000
Hi Dmitry
On Mon, 11 Aug 2025 at 22:17, Шушуев Дмитрий <dimashu@live.ru> wrote (in
the attachment):
> Description:
> Using history builtin with -ps options give unexpected results
> For example after executing:
> dnf -q repoquery --list bash-doc | fold -w $(($COLUMNS/3-1)) | pr
> -3tw $COLUMNS
> if we enter:
> history -p !!
> history 10
> we got:
> 1338 dnf -q repoquery --list bash-doc | fold -w $(($COLUMNS/3-1))
> | pr -3tw $COLUMNS
> 1339 history -p dnf -q repoquery --list bash-doc | fold -w
> $(($COLUMNS/3-1)) | pr -3tw $COLUMNS
> 1340 history 10
> As you see the 'history -p ...' command got in history though it
> doesn't
> This only occur this command 'dnf -q repoquery ...'
The issue is that 'history -p' *retroactively* removes itself from the
in-memory history, which normally does what you expect.
However when it's in a subshell it cannot remove itself from the parent
shell's in-memory history.
Consider:
echo Start
> history -p A B C # (simple command)
> history 1 # outputs « echo Start »
> ( history -p A B C ) # (explicit subshell)
> history 1 # outputs « ( history -p A B C ) »
> history -p A B C | cat # (implicit subshell within a pipeline)
> history 1 # outputs « history -p A B C | cat »
This is in the same broad class of problem as aliases and « shopt -s
extglob »; they affect early stages of parsing (retroactively, in the case
of history -p) but not until they themselves are *executed*, by which time
it may be too late (or in the wrong context) to have the desired effect.
-Martin