Mailinglist Archive: opensuse-project (230 mails)

< Previous Next >
Re: [opensuse-project] openSUSE 2016: taking a picture of openSUSE today
Quoting agustin benito bethencourt <abebe@xxxxxxxx>:

We (openSUSE) have been making a lot of noise about how we have more
users than Fedora. I don't really believe that stat, but I'll go along
with it for now; one thing that we m,ost certainly do NOT have is more
developers using openSUSE compared to Fedora.

This is confirmed by our data. As mentioned by Alberto in his keynote, I think,
we used Fedora as reference because the rest of the distros do not publish
their numbers together with the methodology and the technical info required
for us to compare it.

When we made this analysis we where very carefully redoing all the work to follow the exact way that Fedora used to count installations and downloads. We put extra care here, but there is room for mistakes. The scripts are published so there is a chance to spot the mistake and to change the previous result.

The comparison with Fedora is only because they are the only one that made an _excellent_ job publishing both results and methodology. IMHO we need to push the project to reach the same level of transparency that they showed for this particular case.

Independent of that, data shows a general trend in number of installation and downloads. Of course there is a bias, as Andrew spotted: there are things that are not counted. But again: the exact result is not important, but is the trend that the data shows.

[...]
Not sure I understand what you're trying to get at here. Sure, keep an
eye on trends, but that's all what they are. Don't fixate on it.

The overall number of downloads grow if you take the full time window shown on
the graph. If you count the number of downloads per release, you would see
many variations from release to release.

@Alberto, am I correct? I am sure you can explain better.

If we change the window used to make the data analysis, the final slope change a bit but not the trend. Let me express this in a different way:

We made 3 measures: February, April and July 2013 (a fourth for December), and in the three cases the trend was exactly the same: a bit increase of # of downloads and a bit decrease of # of users. The exact value for the slope was different in every measure, but not much for this small value of slope in comparison with the absolute value of the variable that we are measuring.

A different topic is the explanation of this trend, but for my PoV the real picture is like we have published.



--
To unsubscribe, e-mail: opensuse-project+unsubscribe@xxxxxxxxxxxx
To contact the owner, email: opensuse-project+owner@xxxxxxxxxxxx

< Previous Next >