Bug 342629

Summary:	dev-util/scons-2.0.1 unresponsive to SIGINT
Product:	Gentoo Linux	Reporter:	Martin von Gagern <Martin.vGagern>
Component:	[OLD] Development	Assignee:	Python Gentoo Team <python>
Status:	RESOLVED WORKSFORME
Severity:	normal	CC:	floppym, Marc-Antoine
Priority:	High
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---

Description Martin von Gagern 2010-10-25 12:37:35 UTC

I'm experimenting with the mesa source tree and scons, and found that in my setup, the scons online help fails to terminate, and iterrupting it via Ctrl+C doesn't work either. Killing it via SIGTERM (e.g. with "killall python2" from another terminal) does work.

Steps to reproduce:
1. git clone git://anongit.freedesktop.org/git/mesa/mesa
2. cd mesa
3. git checkout mesa-7.9
4. scons -h

Expected result:
1. scons prints some kind of online help
2. the command exits

Actual result:
1. Scons prints only one line: "Reading SConscript files ..."
2. Scons keeps going forever

Comment 1 Martin von Gagern 2010-10-25 13:27:37 UTC

/usr/portage/dev-util/scons/files/scons-1.2.0-popen.patch seems to be to blame: Removing that patch from the ebuild gives program output and termination as expected. And if I'm fast enough, I can even interrupt the build with Ctrl+C and receive a suitable message before the program terminates.

Therefore this is a Gentoo bug, not an upstream bug, imo.

So you should consider dropping that patch, applying it conditionally where absolutely required, or rework the patch so it doesn't have ill side effects like this.

Comment 2 Mike Gilbert gentoo-dev

2012-02-01 02:40:17 UTC

Do you still have this problem with scons-2.1.0? I see we are still applying the problematic patch, so I'm guessing it is pretty likely.

Just looking to resolve this bug before requesting stabilization of this version.

Comment 3 Mike Gilbert gentoo-dev

2012-02-01 02:42:39 UTC

Also, I can't reproduce the problem using the steps you have outlined, but it could be some kind of race condition.

Comment 4 Martin von Gagern 2012-02-01 08:14:16 UTC

(In reply to comment #3)
> Also, I can't reproduce the problem using the steps you have outlined, but it
> could be some kind of race condition.

Can't reproduce this either. Not even after downgrading to scons 2.0.1. Strange. Feel free to either mark this WORKSFORME or have a closer look at the sources, depending on how much time you have to spare.