[Andrew Kuchling]
> I was curious about how "most active project" is defined, and took a
> look at the SF code. It's computed by the
> util/db_project_weekly_metric.pl script, which sums up a bunch of
> parameters to get a number. The parameters are:
>
> 3 * # of forum postings
> 4 * # of tasks
> 3 * # of bugs
> 10 * # of patches
> 5 * # of support
> 1 * # of CVS commits
> 5 * # of developers
> 5 * # of file releases
> .3 * # of downloads
>
> All of these numbers seem to be counted over the past week.
Since it's Perl (or PHP?), I hope they remembered to change what "past week"
means each week <wink>. Hmm: since they weight patches at 10 times the
value of a commit, I guess they value projects where lots of people are
unhappy but can't agree on what to do about it <log(0.5) wink>.
> I suspect the code has a mathematical bug: the queries use, for
> example, log(4*count(project_task.project_task_id)) for counting
> things. However, IIRC, log ab = log a + log b, so the multipliers
> just add a constant to every project's number, and therefore it's a
> flat sum of all the parameters. IMHO that should be
> 4*log(count(....)), and will report it.
Ha! *Another* bug they wouldn't have had if they had written it in Python
instead! Ya, ya, I'm lying, but if we all agree to say it's the truth,
maybe they'll fall for it.
agreeing-with-your-analysis-but-unsure-why-they're-taking-a-log-
at-all-ly y'rs - tim
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4