On 24 Jul 2015, at 19:17, stepharo <stepharo@free.fr> wrote:
Le 23/7/15 23:49, Yuriy Tymchuk a écrit :
HI,
yes, ban may make sense. I think that âfalse positiveâ is a bit incorrect. For example, consider the âprobably missing yourselfâ rule. Even if you are not missing it, the rule is not wrong. It just warns you that you should be carful with this case, because it is easy to screw up something. Now even while this is not a false positive, you want to skip the rule for this method.
That was the philosophy behind the name. I do not know. We discussed a lot in the past about this vocabulary and it would be good to have one. I do not understand the diff between this is useless and skip. I prefer ban this rule = do not run it false positive = rule is ok but wrong here
Do you mean ban rule completely? On which level? For a developer (machine)? Then again, skip works as former âfalse positiveâ. But in my opinion âfalse positiveâ is not a good name from the logical point of view. You have a rule that detects if you have no #yourself message in the end of a cascade. Now do people really mark this as âfalse positiveâ if the rule detects cascades that have #yourself message in the end? Or they do it if they donât want to see the critic? Then itâs not false positive, then they just want to hide or âskipâ the rule. This is fun part from statistics to name things as false positives. But if we program a rule in such way, that it is fuzzy, I think that itâs not correct to call it as false positive. Maybe itâs critics are useless more times that they are helpful, and if the critic is correct, the issue is not that important. Then probably we should just remove the rule or rewrite it. Now with useless button, the only reason is to collect data on which rules are actually useless to people [1], so we can take a look at them in a first place. It doest not affect the functionality. Uko [1]: http://renraku.inf.usi.ch/rules <http://renraku.inf.usi.ch/rules>
Stef
Uko
On 23 Jul 2015, at 22:29, stepharo <stepharo@free.fr> wrote:
HI yuriy
it would be good that quality assistant uses the same vocabulary than codecritics. for example skip -> ban? marks as false positif?
Stef