GNU bug report logs - #30725
eshell: built-ins do not handle command substitution

Previous Next

Full log

View this message in rfc822 format

From: Jim Porter <jporterbugs <at> gmail.com>
To: yegortimoshenko <at> riseup.net, 30725 <at> debbugs.gnu.org
Subject: bug#30725: eshell: built-ins do not handle command substitution
Date: Mon, 17 Jan 2022 11:41:47 -0800

[Message part 1 (text/plain, inline)]

On 3/5/2018 8:34 PM, Yegor Timoshenko wrote:
> In M-x eshell:
> 
>    $ which echo
>    eshell/echo is a compiled Lisp function in `em-basic.el'.
>    $ which *echo
>    /run/current-system/sw/bin/echo
>    $ echo ${mktemp -d}
>    $ *echo ${mktemp -d}
>    /tmp/tmp.UaiWQ0YPIX

I can see this bug with an even simpler case too: "echo ${*echo hi}".

It turns out that this is because `eshell-invoke-directly' thought that 
the above command was simple enough to, well, invoke directly. However, 
since "${mktemp -d}" or "${*echo hi}" create a subprocess, the command 
needs to be invoked *iteratively* by `eshell-eval-command'. The problem 
was that `eshell-invoke-directly' only checked the top-level command and 
didn't examine subcommands.

Attached is a patch that fixes this, plus a unit test (I've verified 
that the test fails without the patch and passes with it). Note that the 
test *does* rely on the system having an external "echo" command, but I 
think some of the tests in that file already rely on the presence of an 
external "sleep" command, so this should be ok. However, if it causes 
issues on some systems (MS Windows maybe?), just let me know and I can 
try to put a guard around the test so it doesn't run on such systems.

[0001-Consider-subcommands-when-deciding-to-invoke-Eshell-.patch (text/plain, attachment)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.