[Not yet finished] Fix wattmeter probe

[in3otd]

Fix Qucs/qucsator#47 .

There where two main issues:

• the formula used for computing the complex power was wrong:
  this was easy, just use the correct formula for the power
• the voltage noise values from were wrong:
  this required changing the nodes order for the component in the netlist: the probes are a bit of a hack and currently require the voltage sensing to be on the first two nodes, while the wattmeter had the current sensing there.

I also couldn't resist changing a little the probe values output handling to make it feel a little less "hackish" (IMHO).

This goes together with the qucs-test update at Qucs/qucs-test#35 .

[in3otd]

ah, the joys of CI...

• on AppVeyor, pacman doesn't find Qt4
• on Travis, there is some issues with Homebrew and it complains about a missing ccache

any idea is appreciated...

[in3otd]

uh, there's still something wrong: as described in the Qucs docs, "all voltages and currents are peak values" (see pag 13) so the AC power results should be divided by 2. But for DC simulations not.

So the results shown in Qucs/qucsator#47 are wrong, since the 120 V of the source is the peak value and not the RMS as intended in the linked document calculations.

[guitorri]

Concerning CI. MSYS2 seem to have dropped Qt4. One option is to install the official Qt4 package (I used to to it long time ago, it should be on the .appveyor history)... another is to finish the port to Qt5 :). On Travis/OSX you can try to either disable ccache or add it to the brew install list.

[in3otd]

MSYS2 seem to have dropped Qt4

uh, the world is coming to an end 😁
Finishing the Qt5 port would of course be the best solution but, alas, I cannot help much there.

The OSX build log says

Tapped 0 formulae (189 files, 264KB)
Error: Your Homebrew is outdated (need >=2.1.0, found 1.3.5 at /usr/local).
Please run `brew update`.
Error: Kernel.exit

so I guess it's something more than ccache but I know nothing about OSX

[guitorri]

I will try to get Qt5 back on track.
In the mean time, can you rebase?

[in3otd]

Thanks - Qt5 will obviously be the best solution.

As mentioned above, the fix here is not yet finished - I need to separate the DC and AC cases due to the fact that AC uses peak values. I'll rebase anyway in the meantime.

[felix-salfelder]

This is listed as a blocker for 0.0.20. it changes the wattmeter symbol and looks like that will introduce a backwards incompatibility.

better defer to 0.0.21?