FreeBSD branches stats

[Randy Pratt]

On Mon, 23 Oct 2006 13:40:11 +0200 (CEST)
Oliver Fromme <olli at lurza.secnetix.de> wrote:

> Randy Pratt wrote:
>  > I think the page referenced may include "release" numbers from all
>  > the reporting machines regardless of operating system.
>  > 
>  > Better numbers from the relatively small sample can be found at
>  > 
>  >         http://www.bsdstats.org/freebsd/releases.php
> 
> Where is the link to that URL?  I didn't find it anywhere.

Marc (scrappy@) referenced it in a thread on freebsd-questions a
while back but you are correct that it is not linked from the
bsdstats main page.

>  > which tally up to the totals for FreeBSD listed on the main page.
>  > I find the number of 2.x systems a little puzzling though.
> 
> Well, I also have 2.2-stable installed on an old notebook
> that wouldn't reasonably run any newer version (it has 4 MB
> RAM and 120 MB disk, no network, no X).
> 
> However, the number for 2.x systems seems a bit high indeed.
> I think the users who still run 2.x systems tend to tune
> them manually and watch the lists closely for things that
> need to be patched.  Therefore they might have a better
> chance to know about the bsdstats script than people who
> just install a release.  The statistics are certainly skewed
> by the fact that the bsdstats script isn't integrated into
> the standard installation, like other BSD projects did
> meanwhile.  And the numbers are much too low to have much
> of a statistical value so far.

The numbers are definitely much too low to draw any conclusions
but they do raise some questions in my mind.  Had I seen some
number of 3.x systems, the presence of 2.x boxes would seem less
conspicuous.

Speaking for myself, I would not object to bsdstats in the base
system but I'm almost sure it would be off by default.  The problem
of making users aware of its existence would then arise.  Perhaps
more users would participate if they knew it existed.  So far,
bsdstats has only been mentioned in a couple of mailing lists.
The present low numbers of reporting systems just reinforces my
perception that mosts users don't read the mailing lists so even
if it were included in the base system it might not get activated.

All in all, bsdstats could be useful information beyond its original
intent of demonstrating to hardware vendors that a viable market
exists.  I had thought about mentioning these stats during the
4.x EOL bikesheds since the terms "many 4.x" and "a lot of users"
were being tossed around.  The 4.x with 10% total didn't seem to
meet the criteria for "many" to me ;-)  Of course, I didn't bring
it up since the sampling is so small at this point that it would
have only added to the noise on the lists.

Randy



--