Re: developer bug fixing statistics now available

Bertrand Mansion wrote: > > Le 19 déc. 06 à 16:05, Lukas Kahwe Smith a écrit : > >> Olivier Guilyardi wrote: >>> Hi, >>> Martin Jansen wrote: >>>> On Sun Dec 17, 2006 at 12:5507PM -0600, Gregory Beaver wrote: >>>>> Martin Jansen wrote: >>>>>> I suggest we revert the bug fixer rank system until we have >>>>>> figured out >>>>>> how to do it correctly. The current algorithm does not work right >>>>>> and >>>>>> sheds a bad light on people. (Including me, but I don't _really_ >>>>>> mind.) >>>>> Does this look better? >>>>> >>>>> http://pear.php.net/bugs/stats.php#devs >>>> >>>> I'm still wondering if I have "only" fixed 17 bugs so far. I really >>>> can't say, but somehow I feel that this number should be higher. ;-) >>> Why must these very questionable statistics appear on every single >>> package homepage and profile? >>> A design-related "bug" might take me 20 nights thinking and 20 days >>> coding, while a simple "bug" is easily fixed in 5 minutes >>> This sort of statistics will never faithfully represent the complex >>> efforts one put into maintaining a package. >>> But I agree they might be useful especially for developers' >>> _auto-evaluation_. >> >> I think this stat is just a nice feedback when you go on a bug hunting >> spree, but of course every bug is different. Some people write more >> bug free code initially. Also sometimes you work with multiple people >> or you just improve an existing bug report. For example I spend a lot >> of time validating, finding duplicate and patching PEAR::DB bugs, but >> the final resolution was done by Daniel, because I did not feel like I >> had the right to close bugs on that package. >> >> So the key question if its motivates some people while not >> demotivating others. In that case I think its good to keep them. But >> if too many developers feel wronged by the stats, then we should not >> have them. > > I realised only today this "feature" was active on pearweb. I was also a > bit surprised with my rank (#36) since I have been fixing bugs in pear > packages for 6 years now :) Anyway, I assigned a 5-6 bugs I fixed to my > account and was positioned 5 ranks higher already (#31). So I question > the accuracy of this system and wonder if I should assign every bugs I > fixed to myself so my rank could be more in adequation with the actual > work I did ? But if I do so, a new email is sent to I don't know who, > including myself. What's the proper way to do this ? Is there anyway to > automate this ? Hi Bertrand, There have been several auto-enhancements. First, when a dev marks a bug as "closed" it auto-assigns to that dev if the "Assigned" field isn't already marked. This will help with future bugs. I write a script that auto-assigned bugs that were closed using the @php.net address. There are about 800 closed bugs that are not assigned, I have been going through them all by hand(!) to try to weed out the bogus from the closed and figure out to whom credit is due. It's a slow process. I am considering writing a small script that will allow me to quickly review and assign a large batch of bugs, both to do this quickly and to avoid the email "spam" that would result. However, in the mean time, if anyone else is bored and wants something to do, just go to http://pear.php.net/bugs/search.php and search for bugs with status "Closed" then click the "Assigned" column twice, and it will list all unassigned bugs on the first 80 pages of bugs :). Also note I took the stats off the main package page, they only appear on user info and on the bug statistics page, where they are more likely to be helpful than competitive. If there are still objections, let them be known, as I will respond to all helpful criticism (as made evident by previous actions, I hope) Thanks, Greg