On Tue, Dec 13, 2011 at 01:08:49AM +0000, Ben Hutchings wrote: > On Tue, 2011-12-13 at 01:38 +0100, Bill Allombert wrote: > > On Mon, Dec 12, 2011 at 11:49:35PM +0000, Ben Hutchings wrote: > > > Package: popularity-contest > > > Severity: normal > > > > > > The graphs of 'Number of submissions per architecture' at > > > http://popcon.debian.org appears to show that there are about 90,000 > > > current submissions for i386 and amd64, whereas the table to the left > > > shows about 60,000 for each. > > > > > > In general, it seems that the values on the y-axis are about 50% too > > > high. Either that, or there is a subtle distinction in what is being > > > counted in the tables and graphs, that could be made more clear. > > > > As an experiment, I removed the dots, so the line is thinner now. > > Not sure this fix the actual problem, but at least there is less > > clutter. > > Look at the line for i386 on the 12 month graph. It intersects the > y-axis at the label 92,681 and is 1 pixel (about 5%) higher at the right > margin. So this indicates a value greater than 92,681, whereas the > table shows 65,957. > > Then look at the line for armel: it starts above 1448 and rises over the > 12 months, but the table shows 1307. > > Then if you look at the series with the smallest values you can see that > there are steps at the first tick, the second tick, two steps up to the > third tick, four steps up to the fourth tick. > > I think the real values at the y-ticks are 1, 2, 4, ..., 2^i, > 2^(i+1), ...
I reached the same conclusion, and I have updated the website. It should be fine now. Thanks for your report and analysis, -- Bill. <ballo...@debian.org> Imagine a large red swirl here. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
