I am receiving a lot of build failure mails every day for quite a while now ... not finding any time to look at them, I hope you guys are onto those?
~N
Hi Neels,
On Fri, Nov 10, 2017 at 04:02:10PM +0100, Neels Hofmeyr wrote:
I am receiving a lot of build failure mails every day for quite a while now ...
Would be useful for you to state what kind of build failures reported by whom/what. There are gerrit build failurs, osmocom:nightly build failures, osmocom:latest, ...
not finding any time to look at them, I hope you guys are onto those?
I have the feeling this is mostly me at the moment, and I am not entirely happy with the fact that sometimes it appears to me that I have to pick up whatever other people are not looking into. But then, the last week I've been 9 hours ahead of European time zone, so for "nightly" builds I had the "pleasure" of noticing them much sooner than others.
The only long-running OBS build failure that we're seeing is the one related to libosmocore failing "make check" on some OBS buildhosts, but by far not on all of them, due to some CPU flags. It's really sad that this has been going on for months without anyone apparenty trying to properly investigate it (cat /proc/cpuinfo as part of make check) or reproducing this in qemu which can selectively disable any cpu flag.
Regards, Harald
On Sat, Nov 11, 2017 at 10:35:29AM +0900, Harald Welte wrote:
Would be useful for you to state what kind of build failures reported by whom/what. There are gerrit build failurs, osmocom:nightly build failures, osmocom:latest, ...
I'm losing track, all sorts of them, and the amount is such that I have no idea whether they are anticipated / already looked at by someone actively making changes. For the last two weeks my decision was to not spend time there.
The only long-running OBS build failure that we're seeing is the one related to libosmocore failing "make check" on some OBS buildhosts, but by far not on all of them, due to some CPU flags. It's really sad that this has been going on for months without anyone apparenty trying to properly investigate it (cat /proc/cpuinfo as part of make check) or reproducing this in qemu which can selectively disable any cpu flag.
I would appreciate if those people familiar with the CPU features related code would look at it; I would have to start learning about it and the convolutional coding stuff (?) first...
~N
-
Harald Welte -
Neels Hofmeyr