[prev in list] [next in list] [prev in thread] [next in thread]
List: kde-devel
Subject: Improving KDE: Usage, statistics, metrics, etc
From: Chris Arrowood <chris.arrowood () cybertivity ! com>
Date: 2002-12-17 5:39:08
[Download RAW message or body]
I haven't seen this discussed since I joined this list a little bit ago. If it
is an old topic/debate, then I apoligize in advance.
Recently, there was a discussion on this list about whether people used a
particular app or not (aKtion). From the thread, it wasn't apparant if many
people still used the application or not. In fact, it didn't look like there
was a way to tell, except maybe by monitoring the downloads (if it is an app
not bundled w/ KDE).
That got me to thinking... Is there any room for a statistics reporting
mechanism in KDE similar to the one found in Gentoo?
Consider the following:
When a users profile is first created (or when the stats app is installed via
KDE upgrade or as a seperate package) the user is asked if they want to
report stats. If no, the application never runs again (although a shortcut
could exist somewhere to enable it later.) If yes, then periodically stats
would be sent to a central collection area where they are tabulated and made
public.
Potential beneficial information to collect could include:
HARDWARE
- archetecture
- num cpu's
- memory
- etc (sound, video, etc)
SOFTWARE
- OS
- Distribution (if applicable)
- QT verison
- KDE version
- Build Enviroment
- application usage (maybe by extending one of the base classes to increment a
per application counter on exit)
(Please note: I do not know how feasable some of the items above mentinoed
items are. There certainly would need ot be a good deal of collaboration
from experts on each of the various OS's and architectures.)
Even if only a very minimal set of data were collected at first (x86/linux
only) at least we would have the reporting mechanism in place and we could
add modules to detect whatever else as we go.
In all of this email, I assume that we all agree that privacy would be the
first priority. If such an application existed, it should not
gather/distribute any personally identifiable information, beyond a UID of
some sort maybe (again, see Gentoo).
I think some benefits are immediately obvious, while others would probably be
discovered down the road. Off the top of my head:
- by monitoring application usage, we could make intelligent informed
decisions about whether to remove application "X" from the KDE distribution.
We may even realize not only that no one really uses it, but also that most
people use application "Y" instead, which might be a better choice to bundle
w/ KDE
- being able to get a better definition of lowest common denominator WRT
memory, cpu, etc
- a more concrete feel for how many people run KDE
- etc...you get the idea...
I am curious to hear feedback on this... In particular:
- does anyone think this would be worth the effort
- other potential benefits
- cons and potential pitfalls
- individuals interested in contributing
- feedback from someone involved w/ Gentoo's stats package
- possible implementations (esp of the collection mechanism)
Thanks for your patience with this long email. It is late here and I may have
rambled a bit ;-)
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Chris Arrowood
Atlanta, GA, USA
http://arrowood.myip.org/chris/
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic