Dear all, Here are the results from the FEniCS user survey:
https://docs.google.com/forms/d/1bcs2Jwv9pZ8J57rLIUjZBj8BD-_tcUiU11NGX9VkWAs/viewanalytics In short, we have 80% GNU/Linux, 15% Mac and 5% Win, which seems like reasonable figures. There are also some interesting comments at the bottom. Based on these numbers and our web logs for 2013 (or more correctly, one year back from today), I have calculated some approximate download numbers. These are based on finding unique IP addresses / week for downloads of Mac and Win binaries. Summary (mac, win, tot_mac, tot_win, tot) Week: 50.44 63.38 336.266666667 1056.33333333 696.3 Month: 218.573333333 274.646666667 1457.15555556 4577.44444444 3017.3 Year: 2622.88 3295.76 17485.8666667 54929.3333333 36207.6 Exrapolation of Mac downloads gives 17000 downloads / year, whereas extrapolation of Win downloads gives 55000 downloads / year. The truth is likely somewhere in between. The lower figure for Mac is very reasonable since Mac users can also get FEniCS through other sources (building from source and now Macports). Then there's also the issue that these figures are based on a fairly small number of responses to the survey. 5 years ago, when we could track all downloads (before Debian/Ubuntu binaries) we had on the order of 1000 downloads / month, so 50,000 downloads / year is not an unreasonable figure. Then there's the question how downloads relate to users. Is there a rule of thumb for estimating the number of users based on downloads? -- Anders _______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
