Path: csiph.com!goblin1!goblin.stu.neva.ru!usenet.stanford.edu!not-for-mail
From: Roger <rogerx.oss@gmail.com>
Newsgroups: gnu.bash.bug
Subject: Re: Protect Loop Execution with Traps
Date: Wed, 29 Jan 2020 13:28:40 -0500
Lines: 42
Approved: bug-bash@gnu.org
Message-ID: <mailman.56.1580322529.2384.bug-bash@gnu.org>
References: <20200128212551.GD12574@localhost4.local> <20200128020322.GA31704@localhost4.local> <20200128134935.GM1350@eeg.ccf.org> <20200128204932.GB12574@localhost4.local> <20200128210721.GU1350@eeg.ccf.org> <59.1580277932@jinx.noi.kre.to> <20200129143352.GY1350@eeg.ccf.org> <20200129182840.GB5533@localhost4.local>
NNTP-Posting-Host: lists.gnu.org
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
To: bug-bash@gnu.org
Envelope-to: bug-bash@gnu.org
Content-Disposition: inline
In-Reply-To: <20200129143352.GY1350@eeg.ccf.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: list
Xref: csiph.com gnu.bash.bug:15844

> On Wed, Jan 29, 2020 at 09:33:52AM -0500, Greg Wooledge wrote:
>On Wed, Jan 29, 2020 at 01:05:32PM +0700, Robert Elz wrote:
>> and (with all respect to Gred) please avoid archaic uses, and use the
>> commands as they're currently specified, while "trap - INT" and "trap INT"
>> do the same thing, the former is the standard way, similarly for
>> "kill -INT ..." and "kill -s INT ..." the latter is the modern version.
>
>I had <https://www.cons.org/cracauer/sigint.html> open at the time,
>and got the syntax from there.  He actually uses the numeric signal
>specifiers ("trap 2"), which I replaced with the modern standard naming
>strings; but I didn't replace the legacy single-argument form with
>the "-" form.  Bash, of course, accepts both.

The "Proper handling of SIGINT/SIGQUIT" 
(https://www.cons.org/cracauer/sigint.html) looks to be an excellent 
explanation of the problem described here. (I'll add this to my reading list.)

>You're not going to talk me into using "kill -s" any time soon, though.
>I don't think I've *ever* seen anyone use that.

Ditto.  From my experience, "kill -s" is usually used for init scripts or 
debugging difficult bugs with uniquely connected hardware, etc...

>>   | I'm thinking, put "trap INT; kill -INT $$" inside of it's own function,
>
>Yeah, that's pretty common.
>
>sigint_handler() {
>    trap - INT
>    kill -INT $$
>}
>trap sigint_handler INT
>

Extremely much more readable!  My effort at readability was seemingly five 
steps behind your example here.


-- 
Roger
http://rogerx.sdf.org/