On Sat, 19 Sep 2015 12:07:14 +0300, Niko Tyni wrote: > Even prove(1) says '--merge' is risky: > > Caveat: this is a bit of a kludge. In particular note that if > anything that appears on STDERR looks like a test result the test > harness will get confused. Use this option only if you understand > the consequences and can live with the risk. > > I'm inclined to drop the option rather than make it configurable. > This will make the logs a bit harder to read, but should make the > tests more robust.
I think we can live without --merge. (But I also don't object if you prefer a new configuration file :)) > I note that we still need to redirect STDERR, otherwise output on that > will be considered a test failure. Alternatively we could change the test > definitions in the autodep8 package to add the restriction 'allow-stderr'. > Adding a 2>&1 to the 'prove' call in the 'smoke' script seems easier. 'allow-stderr' seems not so ideal to me as it could allow real errors to slip through. Cheers, gregor -- .''`. Homepage: http://info.comodo.priv.at/ - OpenPGP key 0xBB3A68018649AA06 : :' : Debian GNU/Linux user, admin, and developer - https://www.debian.org/ `. `' Member of VIBE!AT & SPI, fellow of the Free Software Foundation Europe `- NP: Willi Resetarits & Stubnblues: I hob mein Kopf valoan
signature.asc
Description: Digital Signature
