[Adding Dererk, my original sponsor, to cc: for his input]

On Sun, Aug 7, 2011 at 10:50 PM, Vincent Bernat <ber...@luffy.cx> wrote:
> I am pretty sorry but providing a binary package that does not depend on
> nvidia blob  does not  make conky  allowed to be  in main.  Policy 2.2.1
> states that  a package in  main "must not  require a package  outside of
> main for compilation or execution  (thus, the package must not declare a
> "Depends", "Recommends",  or "Build-Depends" relationship  on a non-main
> package)".
>
> Therefore, conky should be moved back into contrib.

I was worried that this would be the case, which was why I originally
planned on uploading conky as two separate source packages (see my
earlier reply at [1] for my rationale), but my sponsor convinced me
that this was unnecessary and not preferable, since it would result in
two source packages that would be exactly the same. and that it's
still possible for a source package in main to build-depend on contrib
components and produce binary packages for main. I guess it is indeed
possible since the buildds haven't been complaining about
uninstallable build dependencies...but if it violates Policy, this
shouldn't be at all possible and needs to be fixed.

> Moreover, conky-std was lacking  some functionalities of conky-all. This
> is a  pity that users  that choose to  keep their system with  only free
> stuff do not have access to a complete version of conky just because the
> complete version also has nvidia stuff in it. Users of "main" should not
> be second-class citizens.

That's true; I couldn't think of any solutions earlier however,
besides building yet another binary conky package (conky-all-nvidia?),
but building additional binary conky packages is something I'd rather
avoid. I hadn't thought about the possible solution you mention below
though. ;)

> It seems that nvidia.c could  be provided into an alternate version that
> uses nvidia-settings if installed. From:
>  https://wiki.archlinux.org/index.php/NVIDIA#Method_1_-_nvidia-settings
>
> $ nvidia-settings -q gpucoretemp -t
> 41
>
> If you can  provide the name of  other settings, I could come  up with a
> patch to solve this.

An excerpt from [2] (scroll down to "nvidia"):

Nvidia graficcard support for the XNVCtrl library. Each option can be
shortened to the least significant part. Temperatures are printed as
float, all other values as integer.
threshold - The thresholdtemperature at which the gpu slows down
temp - Gives the gpu current temperature
ambient - Gives current air temperature near GPU case
gpufreq - Gives the current gpu frequency
memfreq - Gives the current mem frequency
imagequality - Which imagequality should be chosen by OpenGL applications

Example Conky output:

${nvidia threshold} -> "127" (in degrees celsius)
${nvidia temp} -> "40" (also in degrees celsius)
${nvidia ambient} -> "N/A" ("ambient" doesn't seem to be working for
me for some reason)
${nvidia gpufreq} -> "169" (in MHz)
${nvidia memfreq} -> "100" (also in MHz)
${nvidia imagequality} -> "1" (valid values are from 0 to 3, 0 = high
quality, 3 = high performance)

This is what I've managed to deduce so far:
$ nvidia-settings -q GPUCurrentClockFreqs -t
169,100
(first value = ${nvidia gpufreq}, second value = ${nvidia memfreq})

$ nvidia-settings -q OpenGLImageSettings -t
1
(= output of ${nvidia imagequality})

I've also dumped the output of "nvidia-settings -q all" into a
pastebin [3], if it helps any. I'm not sure about ${nvidia threshold}
or ${nvidia ambient} though, sorry.

If you'd like to write a patch to implement this, I'd be glad to test it!

- Vincent

[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=579102#47
[2] http://conky.sourceforge.net/variables.html
[3] http://pastebin.com/KNGHiCVm



--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to