| 1 |
On Sat, Apr 3, 2010 at 7:59 AM, Tobias Scherbaum <dertobi123@g.o> wrote: |
| 2 |
> Am Samstag, den 03.04.2010, 15:40 +0100 schrieb Roy Bamford: |
| 3 |
>> First, we need some metrics - the first step to controlling anything is |
| 4 |
>> to measure it. |
| 5 |
> |
| 6 |
> So, how do you want to measure those metrics? I for one can't think of a |
| 7 |
> useful algorithm which helps to identify understaffed or orphaned areas. |
| 8 |
> Sure, one might take a look at the number of packages compared with open |
| 9 |
> bugs for example - but in the end that still won't give you some useful |
| 10 |
> metrics. |
| 11 |
|
| 12 |
When I was a treecleaner I tended to look at a few things; note that |
| 13 |
because we enforce very little in the tree these are basically just a |
| 14 |
set of heuristics. |
| 15 |
|
| 16 |
- metadata.xml: how many packages are maintainer-{needed,wanted}. |
| 17 |
Does not apply to all herds because some herds fix anything in their |
| 18 |
herd. |
| 19 |
- date of last commit: Gentoo is fast moving and packages that |
| 20 |
haven't had commits since 200{4,5,6} are probably old, unmaintained |
| 21 |
and may not even compile or run. |
| 22 |
- date of last listed maintainer commit versus last commit: |
| 23 |
Basically if the maintainer hasn't touched the ebuild in a while but |
| 24 |
someone else (herd members?) have, the metadata.xml is probably out of |
| 25 |
date. |
| 26 |
|
| 27 |
The above are all pretty easy to do with the data in the tree. Some |
| 28 |
other useful ideas might be: |
| 29 |
- compare open bugs for the package, when was the last bug for a |
| 30 |
package closed (bugs data kinda sucks for this) |
| 31 |
- for a given package in a herd, check the version in the tree |
| 32 |
against freshmeat or similar to see how far behind it is (I think |
| 33 |
someone wrote something for this already, exherbo?) |
| 34 |
- check imlate to see if keywording is behind (is the maintainer |
| 35 |
filing stablereqs?) |
| 36 |
|
| 37 |
Metrics do not have to be perfect (they never are...) but they may |
| 38 |
shine some light on some areas of the tree that need staff. |
| 39 |
|
| 40 |
-A |
| 41 |
|
| 42 |
|
| 43 |
> |
| 44 |
> If someone has a feeling somewhere helping hands are missing or an area |
| 45 |
> is orphaned - that's the best "metrics" we can get. |
| 46 |
> |
| 47 |
> - Tobias |
| 48 |
> |
| 49 |
> -- |
| 50 |
> Praxisbuch Nagios |
| 51 |
> http://www.oreilly.de/catalog/pbnagiosger/ |
| 52 |
> |
| 53 |
> https://www.xing.com/profile/Tobias_Scherbaum |
| 54 |
> |
Archives