While reading LpSolit’s article about the number of people fixing bugs on Bugzilla I thought that I should try to do the same for Calendar. Unfortunately this didn’t really work as a lot of our fixed bugs were never assigned to the person who developed the fix and that completely screws our numbers.
So I thought of a second-best option: The overall number of bugs fixed per quarter for the last five years (click to enlarge):
What the numbers (and especially the moving average) clearly show is that we’ve been picking up some serious steam and are more active than ever.
A similar but cumulative graph can be obtained from Bugzilla itself:
https://bugzilla.mozilla.org/reports.cgi?product=Calendar&datasets=RESOLVED%3A&datasets=VERIFIED%3A
Note that the cumulative graph also includes bugs with a resolution different from FIXED. ;)
You could try this one instead:
https://bugzilla.mozilla.org/reports.cgi?product=Calendar&datasets=FIXED%3A
or, if you want to see all resolutions:
https://bugzilla.mozilla.org/reports.cgi?product=Calendar&datasets=FIXED%3A&datasets=INVALID%3A&datasets=WONTFIX%3A&datasets=DUPLICATE%3A&datasets=WORKSFORME%3A&datasets=INCOMPLETE%3A&datasets=EXPIRED%3A
You suddenly realize that more than 50% of bugs filed are invalid/dupes.
Breaking down the last 2 years a bit more closely shows some other interesting trends. Most notably, the huge impact that the 0.5 (May/June ’07) and 0.7 (Oct ’07) freezes had on FIXED rates. Oddly, this didn’t appear for the 0.3 release (Sept/Oct ’06).
http://img340.imageshack.us/my.php?image=calbugsav0.png
Note that the October ’07 number was pro-rated (10*(20/31)).
Oops, I meant 10*(31/20). The correct number was used in the chart.
jminta, I think that the impression that you give is misleading.
For example during the 0.5 freeze period we had 30 bugs that got patches and reviews that were only checked in after 0.5 was released [1]. And that screwed our numbers.
The same will be true for 0.7 where we already have 23 bugs that wait to be checked in after the release [2].
Lastly, I think our numbers differ a little bit. I counted all the bugs that were fixed during a period, that are still marked as FIXED. Looking at your numbers, you are counting all the bugs that were marked FIXED during a time period, but those could have been reopened or moved to other resolutions (like WFM).
[1] http://wiki.mozilla.org/Calendar:Post_0.5_Checkin
[2] http://tinyurl.com/yodjf9
Simon,
I did not intend to give the impression that the charts show 0.5/0.7 freezes had an impact on the rate of calendar development. I was merely trying to provide a descriptive explanation for the 2 most obvious trends in the data. Interestingly, notice that even with the post-0.5 bump in July, August and Sept saw even *more* FIXED bugs.
I disagree with your second criticism however. I intended to include only bugs currently marked FIXED. (I had the Resolution:FIXED selection made.) I also think it’s quite rare that the FIXED bug gets moved to WFM. Nonetheless, if your chart looks different, let me know.
Finally, if you’re looking for more fun numbers to crunch, I’d be curious at what Bonsai says. Obviously the correlation between FIXED bugs and productivity is only loose, because it doesn’t distinguish between bugs that take 5 minutes and bugs that take 5 days to fix. (Both charts suffer from this.) Maybe a Bonsai analysis of lines-changed by quarter/month would get closer to this. It would still fail to distinguish between search-and-replace and “oh my gosh I just wrote the today pane” checkins though.
Joey, I just compared some the numbers for May, June and July 2007. I counted
– all the bugs that were marked FIXED during that period
– all the bugs that were marked fixed then and are still marked as FIXED
Here are my numbers:
May 20 15
June 35 28
July 80 78
Looking at your graph it seemed to me that you counted the bugs from the first scenario, although your May number does not fully correspond to mine.
Regarding the FIXED vs. WFM thing. That happens quite often as new bug reporters do not really distinguish between FIXED and WFM and mark their reported bugs as FIXED if they can no longer reproduce it.
Finally, I agree that the bonsai statistics could be an interesting exercise. Just for the fun of it I tried to find out, who fixed most of the bugs in Calendar since 2003. I come up as number 6, shortly after mickey and mvl and even before daniel and philipp. And I don’t think that my contributions can be compared with theirs.
Just FYI, you come up on 2nd place behind Mostafah ;)
I can post those stats as well if you want me to.
Which leaves us with the question: Who is number three???
:-)