TOPLIST / FreeBSD usage per 1M inhabitants

Please elaborate.
See the post from ralphbsz above.

You mean the source code of the website? Is the raw data freely accessible?
No idea. I tried to investigate whether it would be possible to extract raw historical data through SQL injections (yep, see the source for that error page) to crosscheck numbers on the site (what "this month" even means? last 30 days? 1st day of the current month until now? wtf?), but that frontend is so anemic, it's completely ridiculous.
 
None of us working on improving it though.
I just meant that we can take that shell script and make it better if needed and set up another website to show the results. The script is simple and straightforward, maybe using HTTPS would be a a good idea to avoid somebody spoofing the reports.
 
So, in a nutshell, it is "useless" because it's not widely used. :-/
No, I said "worse than useless", because those limitations are not mentioned anywhere on bsdstats.org, thus it's misleading. Moreover, as you can tell from my post above, I have zero trust that the site doesn't mangle data it actually receives. We really need the source code.

I'd be interested in download stats for official mirrors instead, that at least would make some sense.
 
So, in a nutshell, it is "useless" because it's not widely used. :-/

Just saying if there's causality, it's the other way around. Then, you'd still have to explain what makes it useless.
My personal opinion is that there is no big problem with this website or the script. The only problem is that this is not widely used. One cannot expect to create a new statistics mechanism and this new mechanism to become instantly widely used. At least as long as the usage is voluntary.

Another thing is that the website provides an e-mail address of the maintainer. If anybody has a very strong concerns or suggestions, maybe try to send a private e-mail. The port seems also to be maintained.

... and maybe the author cooperates and some people here can even offer a helping hand to fix the issues. The ticket is open in Bugzilla...
 
What about contacting ArsTechnica?
Maybe they should also report on BSDstats, so that the pros and cons of its behaviour can be discussed in the larger public?

Edit 2:
Before I contact ArsTechnica, I'll email Colin Percival (FreeBSD security officer) and Ed Maste asking for comment and remedial action.
I hope this will be sufficient so that the issue can get solved without negative publicity for our beloved FreeBSD.

Edit 3:
FreeBSD Project Manager, FreeBSD Security Officer and FreeBSD Security Team have been informed about the issue.
I hope it will be remedied quickly.
You chose a disgraceful approach to getting this issue addressed. Your bug report post, your email to the freebsd-security mailing list, and this post all threaten "negative publicity," which shows you have the emotional maturity of a child. You eventually figured out how adults approach these issues (emailing the relevant FreeBSD team members), but not before embarrassing yourself.

I hope you learn from this.
 
- add a post-upgrade message informing the user it's now opt-in and may need enabled

Which means it's broken upon arrival and you have to agree to let it perform its intended function by enabling it to do so.
 
- add a post-upgrade message informing the user it's now opt-in and may need enabled

Which means it's broken upon arrival and you have to agree to let it perform its intended function by enabling it to do so.
After upgrade it just crashed like this:

Code:
bsdstats-send
/usr/local/etc/periodic/monthly/300.statistics: ${$m...}: Bad substitution

Ran portdowngrade sysutils/bsdstats r537854 and reinstalled to get it working again...
 
- add a post-upgrade message informing the user it's now opt-in and may need enabled

Which means it's broken upon arrival and you have to agree to let it perform its intended function by enabling it to do so.
It's no more broken than any other package installed from ports that requires you to set a variable in periodic.conf or rc.conf to enable it's functionality or explicity run the program.

The majority of ports (I haven't checked every port) require some form of user intervention post install to run them, wether that be enabling them in rc.conf/periodic.conf or explicitly running the application. So it may not be detailed explicitly in FreeBSD documentation but there is an expectation that pkg install/ make install will install a port and nothing else.

After upgrade it just crashed like this:

Code:
bsdstats-send
/usr/local/etc/periodic/monthly/300.statistics: ${$m...}: Bad substitution

Ran portdowngrade sysutils/bsdstats r537854 and reinstalled to get it working again...
Apologies for that, there was a typo that was missed in the original patch, I've submitted a new bug report (PR 255631) to get that corrected.
 
bsdstats is a small way to see who has what

the marketing and bekantheit bsd does not get to the end customers

very sad
 
The script is simple and straightforward, maybe using HTTPS would be a a good idea to avoid somebody spoofing the reports.

What I think is worse than that is that an eavesdropper or MITM can get detailed information about the software that's installed, including full package versions and the output of uname -r.
 
The data is certainly wrong, because it would suggest that Finland has ~13 FreeBSD users. But e.g. a local Twitter group already has 35 followers so certainly ~13 users for the entire country is suspicious.
 
There are more suspicious things in life. These groups here are ok. They don't ask for money. Do you know this or that . Let me respond.
 
Back
Top