I’m trying to build a database of what packages users have installed on their machines. I’ve been wanting to do this for a while. I think it would help out in a number of places, like before considering what packages to remove.
Here’s the webpage which gives all the details: http://stats.larrythecow.org/
If you just want to cut to the chase though, and upload your statistics, you’ll need curl + eix or portage-utils.
Here’s the quicky:
curl -d "arch=$(uname -m)&packages=$(eix -I ––only-names)" http://stats.larrythecow.org/
or
curl -d "arch=$(uname -m)&packages=$(qlist -CI)" http://stats.larrythecow.org/
Thanks guys. I’ll make a prettier page once I have some actual data and something worth sharing.
Edit: Sorry about some of the characters, wordpress was messing them up.
Nice idea!
This will get some use full statistics I think. For example which packages to focus on when stabilizing etc. 🙂
Nice work!
That’s a pretty interesting idea, actually, since it would also show what useful things there are that maybe I should install 🙂
I would also suggest maybe also getting statistics on the packages in the world file (and custom sets, I guess) so that it would be possible to see what people are actually installing (not just the stuff pulled in). That way, it would be possible to differentiate between what people are actually using and stuff that’s just pulled in (as a default use-flag dependency, for example).
Cool idea. Might want to update your page. For the first example, there should be two dashes in –only-names rather than just one. Unless it was changed in another version of eix. Interesting project.
@gregf,
Thanks, it was silly WordPress messing it up.
I like it, submitted all of my gentoo boxes to it.
After fighting with the funky quotes when I copy-pasted the commands that is 😉
So that’s 15 more results submitted from the servers I manage.
I wonder, what are you doing to avoid duplicates/double-posts/abuse?
lol, curl not installed;)
The output of eix differs from the one of qlist. For me is is as follows…
# qlist -CI | wc
1843 1843 36639
# eix -I –only-names | wc
1632 1632 32526
I think, the output of qlist is more accurate, as is reflects all real installed packages. eix only shown those package, it knows ebuilds for. If you have some packages installed from an overlay and later remove the overlay, eix will miss those packages.
Great to see that someone implements this kind of thing. Was not there a Google Summer of Code project about this a couple of years before? What happened with it?
About time!!! I’ve been looking for something like that for ages!!!
Debian made me jealous: http://popcon.debian.org/
woho. +3 machines (=
Hope you’re aware that the data won’t be very useful in the long run without client tracking. One client can upload his data multiple times and screw up the results. Also you only track what people had installed at one point in time, so packages installed once will be stuck in the database forever, even if they’re long gone from the tree and user systems.
You might want to look at the popularity-contest package from Debian.
@Tobias, eix just lists each slot once. It should be the same if you pipe it through uniq.
@genone,
True of course, I don’t have any long term plans right now though. I just got tired of having it be a wish list item of mine, and slapped it together. I’m impressed it actually went so quickly. We can always build on it later, but for now I’d just like to gauge user activity — if people actually participate, then it’ll be worth expanding. If not, oh well. Nothing venture, nothing gained. 🙂
nice!
Please also record USE flags, I think this is important to know too.
Hey, great idea, I wouldn’t mind if you submit list of my packages during sync, ofcourse if users can turn this on/off in config. It would be great if that way devs can see what are the most common packages users install that are not in portage and add em 😉
Something like this should get into Portage, just like Debian and Ubuntu have. It really provides good stats than can be used to make important decisions.
Perhaps you should get in touch with the guy from portagefilelist.de. He has collected alot of stats. A combination of your two project would be more powerfull for both sides.
Just uploaded my list using the qlist variant of your command above. It’s from my media server.
>I’ll make a prettier page once I have some actual data
>and something worth sharing.
Maybe over 1100 lists? So…
The upload script responds with:
> Thanks 🙂 We now have 1418 entries.
while the page showed that amount a while ago (after upload it shows 1 more). So, it looks like script displays count before adding new data.
BTW. WP fails with polish utf-8 chars:
> WordPress database error: [Illegal mix of collations (latin1_swedish_ci,IMPLICIT) and (utf8_general_ci,COERCIBLE) for operation ‘=’]