GNU bug report logs - #59103
29.0.50; esh-var-tests intermittently fails on macOS 12.6.1

Previous Next

Full log

Message #8 received at 59103 <at> debbugs.gnu.org (full text, mbox):

From: Jim Porter <jporterbugs <at> gmail.com>
To: Stefan Kangas <stefankangas <at> gmail.com>, 59103 <at> debbugs.gnu.org
Subject: Re: bug#59103: 29.0.50; esh-var-tests intermittently fails on macOS
 12.6.1
Date: Mon, 7 Nov 2022 21:39:57 -0800

On 11/7/2022 3:09 AM, Stefan Kangas wrote:
> Intermittently failing test on macOS 12.6.1, see below output from
> "make -j5 check":
> 
[snip]
> They do not fail every time: sometimes neither fails, sometimes one
> does and sometimes the other.  Only once did both fail at the same
> time, after several runs of "make check".

Both of these failures look like the ${subcommand} didn't see any output 
by the time Eshell expanded it into the top-level command. My guess is 
that something is going awry in 'eshell-sentinel'. I haven't been able 
to reproduce this locally (on a GNU/Linux system), but since it seems 
you *can* reproduce it, you could try the following:

1. Check if 'eshell-output-object' ever gets called from within 
'eshell-sentinel' or from the 'finish-io' timer defined there

2. If so, try moving the 'eshell-kill-process-function' call at the end 
of 'eshell-sentinel' (and maybe the 'when-let' block immediately above 
it) inside 'finish-io'

My reasoning here is that I think 'eshell-kill-process-function' is what 
will signal to the rest of Eshell that the process is done, but if the 
process still has output pending (i.e. 'finish-io' gets called from a 
timer to do its thing), then that pending output would get lost.

Hopefully that at least points you in the right direction...

Previous Next

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