On 04/08/2015 05:51 PM, Justin Clift wrote:
On 7 Apr 2015, at 11:21, Vijay Bellur <vbellur@xxxxxxxxxx> wrote:
<snip>
3. Spurious regression tests listed in [3] to be fixed.
To not impede the review & merge workflow on release-3.7/master, I plan to drop those test units which still cause
spurious failures by the time we branch release-3.7.
Thinking about this more... this feels like the wrong approach.
The spurious failures seem to be caused reasonably often by
race conditions in our code and similar.
Dropping the unfixed spurious tests feels like sweeping the
harder/trickier problems under the rug, which means they'll
need to be found and fixed later anyway.
I don't think that is the intention. Rather than sweeping the
harder/tricker problems away, at least my original intent was to spur
ourselves into action to clean up the frequently failing tests.
And yeah, this could delay the release a bit. Personally, I'm
ok with that. It's not going to make our release of lower
quality, and people not involved in spurious failure fixing
are still able to do dev work on master.
I am open to this. We need not stop branching but can block the release
till the listed spurious regression failures are fixed.
Thanks,
Vijay
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-devel
