Iiro Laiho wrote: > No. I am just describing the current QA procedure. The +1 comments from > three people are seen as some kind of proof of stability of the update, > and those commenters don't go through any screening of any manner. "Works > for me" really is a common expression in those comments. That we often > don't know anything about those random people that happen to comment the > update and still trust their word *is* a part of the problem. Indeed, that imaginary "karma" number with no real-world meaning is just broken. Requiring updates to have that number above a certain threshold serves no practical purpose whatsoever, because: * as you pointed out, we have no idea whatsoever who those people are, or even whether they are real persons, * automatically summing up unrelated +1 and -1 comments with no human looking at the contents does not work – 4 "+1 works for me" and 1 "-1 destroys all my data" disastrously sum up to "+3". Even more silly is the way use of autokarma is encouraged, which even eliminates the last chance of screening by a thinking being, by: * enabling it by default, and * keeping issues like: https://github.com/fedora-infra/bodhi/issues/1032 https://github.com/fedora-infra/bodhi/issues/1033 unfixed for weeks! (And before that, there was: https://github.com/fedora-infra/bodhi/issues/772 whose fix was not deployed for months, and then turned out to not work, also because an overengineered solution was chosen (and implemented in an incomplete way) instead of the straightforward one that I proposed and that would have been totally equivalent policy-wise.) Manual push to stable based on karma has now been broken for MONTHS and nobody cares! And when I complain on IRC, everyone says they just enable autokarma, which is a horrible "solution", because autokarma is a totally non-workable concept. When will Fedora stop giving meaning to a totally meaningless integer aggregating, without any weighting, the feedback of totally untrusted users? A computer CANNOT adequately process user feedback, only a human can. Kevin Kofler _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
