[gentoo-portage-dev] [PATCH] SyncManager: redirect command stderr to stdout (bug 566132)

[gentoo-portage-dev] [PATCH] SyncManager: redirect command stderr to stdout (bug 566132)

[Zac Medico]

X-Gentoo-Bug: 566132
X-Gentoo-Bug-URL: https://bugs.gentoo.org/show_bug.cgi?id=566132
---
 pym/portage/sync/controller.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/pym/portage/sync/controller.py b/pym/portage/sync/controller.py
index 4595293..e71ba67 100644
--- a/pym/portage/sync/controller.py
+++ b/pym/portage/sync/controller.py
@@ -218,6 +218,13 @@ class SyncManager(object):
 
 		self.usersync_uid = None
 		spawn_kwargs = {}
+		# Redirect command stderr to stdout, in order to prevent
+		# spurious cron job emails (bug 566132).
+		spawn_kwargs["fd_pipes"] = {
+			0: sys.__stdin__.fileno(),
+			1: sys.__stdout__.fileno(),
+			2: sys.__stdout__.fileno()
+		}
 		spawn_kwargs["env"] = self.settings.environ()
 		if repo.sync_user is not None:
 			def get_sync_user_data(sync_user):
-- 
2.4.9

Re: [gentoo-portage-dev] [PATCH] SyncManager: redirect command stderr to stdout (bug 566132)

[Alexander Berntsen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Sorry for not being up to date what this does, but the bug complains
specifically about GitHub mirrors, while the patch itself seems far
more invasive. Is it the case that they should *never* go to STDERR?

- -- 
Alexander
bernalex@gentoo.org
https://secure.plaimi.net/~alexander
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQIcBAEBCgAGBQJWUuzpAAoJENQqWdRUGk8BMdQQAKACO5bGwnXxbyNRuh/KzIiH
e38mzzvxDibJTTz32XSsOn1w9StZkSWAcDwT8Ef4QGeQJxmLTXUW+5eUsAVoaNxy
VY81iQbDmcSf+erljYEuUBBOFDMORYGwhSROdvvLE/0d3LFP6AurWx9a4kb229j7
DB8HmpjXkkpT65YgJvp2vD51x+K5HsDOKTObBcDreiea21u2X6DNcGbVOK35IXs3
fbTovSrL39qTk8zPSvuwePHlfWelsTCmj/+2EmLVJMTem/cTo0uxCfGCOmwALusk
FI4YcV1zf1cMrL1tqU9B7bYSH8oci4Rfs80IanieDwpd59V32xQ8ckyOP8ME5gG/
xbqipfwbWJyyUFErL7BZE9/H90tib9uofR9mxP9BylXjsDMbu/sC7OA2ElBNVilp
TJGvmRb3jdIWsSddNsOgdN2XofR20G5Ow/NKEvpueM42Pk4s6lqYwZODmsU5kOsS
wSw6ErQO6RCKfu0iQhQo3VS4EjP95sVl21nXtJk6v8AkNwNkU1TNZ8f1Ds5CaALR
79XpFL3JQy/zwd0uMY7kOjgezJJeu4nhfSz2k1ZDNLHokTcqhqR4WdXqzkWA2iZq
Y6p6CLWY57TIiih0ASt+NSfrkRl1628irZyVcyoVKDpaWCVcWq+G1sAl4pmAXpXV
nVhyL0zqartr+TNlFz2N
=g6l5
-----END PGP SIGNATURE-----

Re: [gentoo-portage-dev] [PATCH] SyncManager: redirect command stderr to stdout (bug 566132)

[Zac Medico]

On 11/23/2015 02:39 AM, Alexander Berntsen wrote:
> Sorry for not being up to date what this does, but the bug complains
> specifically about GitHub mirrors, while the patch itself seems far
> more invasive. Is it the case that they should *never* go to STDERR?

Since the sync modules can call a variety of different commands, it's
hard to predict which ones will unnecessarily write to stderr. So, the
assumption here is that the process exit status is a more reliable
indicator for errors. For example, the python code in GitSync will write
an error message to stderr if the git process exits with non-zero status.
-- 
Thanks,
Zac

Re: [gentoo-portage-dev] [PATCH] SyncManager: redirect command stderr to stdout (bug 566132)

[Alexander Berntsen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 23/11/15 17:46, Zac Medico wrote:
> Since the sync modules can call a variety of different commands,
> it's hard to predict which ones will unnecessarily write to stderr.
> So, the assumption here is that the process exit status is a more
> reliable indicator for errors. For example, the python code in
> GitSync will write an error message to stderr if the git process
> exits with non-zero status.
OK, then ACK, I suppose.

- -- 
Alexander
bernalex@gentoo.org
https://secure.plaimi.net/~alexander
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=cvLG
-----END PGP SIGNATURE-----