Re: Statistics on non-free usage
On Thu, Jan 08, 2004 at 02:51:36PM -0600, John Goerzen wrote:
> On Thu, Jan 08, 2004 at 01:00:12PM -0600, Steve Langasek wrote:
> > > >From the data, we can see that:
> >
> > > * The 5 most popular packages in non-free are acroread (18% regular
> > > use), unrar (14%), j2re1.4 (11%), and rar (10%).
> >
> > acroread is no longer distributable (or distributed), so should probably
> > be excluded from any analysis.
> >
> > Also, are any of the java packages actually distributed by Debian? I
> > thought there were legal issues that prevented even non-free
> > distribution (though j2re/sdk packages are available elsewhere).
>
> Excellent points. No, acroread is not in non-free. j2re1.4 also is
> not, nor is j2dsk1.4 or, in fact, any Java newer than 1.1. I dare say
> that Java 1.1 in non-free is about the same usefulness as Kaffe for
> today's programs.
Yep, maybe it would be a good candidate for removal.
> So, we have a situation where the #1 and #3 packages installed from
> "non-free" on people's systems are not actually present in Debian's
> non-free (any more). Also, no version of Java later than 1.1 is
> present.
yep, but these are also the most high profile packages. I guess the
situation is different for other packages in non-free.
Also, installing java stuff from third party sources is a pain. See for
example the problem with mozilla-cvs and mozilla-snapshot, which you
have to hand fix in the postinst. Also, there is no 1.4 .deb for powerpc
for example.
> > > * In main, gs has 42%, xpdf-reader 26%, gv 20%. tar was at 87% and
> > > unzip at 49%.
> >
> > Of course, tar and unzip are no substitute for unrar.
>
> It, of course, depends on what you're doing, but yes, I realize that. I
> just tried to show a smattering of similar programs so people can
> compare.
>
> I was actually surprised at the popularity of {un}rar. I rarely see RAR
> files used anywhere.
Maybe some dependency ?
> > Interesting statistics. Thanks for doing this, John.
>
> Glad to do it.
But as you said, it doesn't really prove anything, only that the people
using popularity contest don't really use these non-free packages much.
What about all those who don't run popularity contest, or those who are
offline ? What about monitoring BTS traffic for those packages ?
Friendly,
Sven Luther
Reply to: