[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Signal handling by readline
From: |
Elad Lahav |
Subject: |
Re: Signal handling by readline |
Date: |
Mon, 30 Sep 2024 16:43:37 -0400 |
Thanks for taking the time to respond.
On Mon, Sep 30, 2024 at 4:19 PM Chet Ramey <chet.ramey@case.edu> wrote:
> Bash isn't the only application using readline. Most applications using
> it don't do anything special with the job control signals. Bash, being
> a job control shell, is of course the exception.
Understood.
> There is no window, at least in versions of readline released in the
> last decade. rl_set_signals() blocks the signals it modifies before it
> changes any of them, and unblocks them after changing all of the
> dispositions.
Ah, I missed that the functions assume that the signal is blocked.
> Applications that don't want readline to do this can disable it, at a
> cost.
But it is still the case that bash makes many redundant system calls
on every line entered. Is setting rl_catch_signals to 0 in bash going
to have any adverse effects? What about changing it to a signal mask
that the application can set, such that readline knows which signals
it should touch and which ones to leave alone?
If there is no race condition, though, then this is purely a
performance issue for something that is not performance-sensitive.
--Elad