| 1 |
On Mon, 23 Apr 2018 22:30:21 +0100 |
| 2 |
"M. J. Everitt" <m.j.everitt@×××.org> wrote: |
| 3 |
|
| 4 |
> I'm inclined to support the idea that <something> is marginally better |
| 5 |
> than <nothing> and could be used as some form of evidence should there |
| 6 |
> be need to substantiate a claim either way, and make it much less |
| 7 |
> subjective ... |
| 8 |
> |
| 9 |
> Not looking for a perfect solution here, that IS impossible, just |
| 10 |
> something that would aid metrics... |
| 11 |
|
| 12 |
Agreed. Its useful in one direction, to say "Look, they're active, proof!", |
| 13 |
but can't be used to say "Look, they're INACTIVE". |
| 14 |
|
| 15 |
Similar to how a Bloom Filter can be used to say: |
| 16 |
|
| 17 |
"We definitely did not see this item before" |
| 18 |
|
| 19 |
But can't be used to say |
| 20 |
|
| 21 |
"We definitely saw this item before" |
| 22 |
|
| 23 |
Subsequently, an automated "you're inactive, so drop your commit |
| 24 |
rights" bot can't exist, as it can only determine if you're factually |
| 25 |
active, not factually inactive. |
| 26 |
|
| 27 |
|
| 28 |
It could at best aggregate these sources and produce a list of gentoo |
| 29 |
devs that are /possibly/ inactive, but not /certainly/ inactive. |
| 30 |
|
| 31 |
And humans can then scour that list and attempt to verify the activity |
| 32 |
status. |
Archives