Fix double call to receiver::set_rf_freq, may fix #1129

[0penBrain]

Also commented at conf loading because same thing arises there.

[argilo]

may fix #1129

Before merging a change intended to fix a bug, I'd want to understand the root cause of the bug in question, and have some confirmation that the bug is actually fixed.

I have my doubts that the duplicate setFrequency calls are actually causing #1129, because it wouldn't explain why it only happens when tuned to a bookmarked frequency. And even if duplicate setFrequency calls were triggering the issue, removing duplicate calls here wouldn't fix the root cause (which presumably would be in in rtl-sdr or gr-osmosdr). There are other ways of triggering rapid setFrequency calls, like dragging the mouse across the plotter.

[0penBrain]

Before merging a change intended to fix a bug, I'd want to understand the root cause of the bug in question, and have some confirmation that the bug is actually fixed.

I'm sorry I can't help because I'm not reproducing here. But at least guys reproducing can get this branch and test. 😉

Also notice that even if it doesn't fix the bug, it still is more proper implementation. 😄

I have my doubts that the duplicate setFrequency calls are actually causing #1129, because it wouldn't explain why it only happens when tuned to a bookmarked frequency.

I share this but at some point you have to try something. 😄 Also notice in the bug discussion, no one formally said "it happens only when I'm on a bookmark and activate another bookmark, but doesn't happen if I'm on an arbitrary frequency and activate a bookmark". 😉

And even if duplicate setFrequency calls were triggering the issue, removing duplicate calls here wouldn't fix the root cause (which presumably would be in in rtl-sdr or gr-osmosdr). There are other ways of triggering rapid setFrequency calls, like dragging the mouse across the plotter.

The time frame of the dup call wouldn't be exactly the same, as when dragging the mouse it would happen on (at least) 2 different runs of Qt event loop. Here it happen in the same run (guess so, a bit unsure about Qt signal/slot).

[argilo]

it still is more proper implementation

Sure, if there are reasons other than #1129 for making this change, we can make it. I'll need to review to make sure there aren't any unintended side effects.

you have to try something

Yeah, experiments to get to the bottom of #1129 are a good idea. Another thing to try might be intentionally stressing the RTL-SDR with very frequent setFrequency calls.

[willcode]

Blocking the signal in freqctrl::setFrequency() to avoid the loop would be better, architecturally. There are probably multiple places where we depend on a signal being routed through a GUI element, but it's not terribly modular that way.

The problem here is that setFrequency() is also used internally in the widget. That would have to be changed, so it was obvious that the change was due to a signal.

[0penBrain]

Blocking the signal in freqctrl::setFrequency() to avoid the loop would be better, architecturally.

Yes. Thought to that but this is far more changing and testing, because the signal is actually connected to 3 or 4 slots in MainApplication. Would deserve an architectural decision first.

The main problem IMHO is that 2 functions are calling for a (hardware) frequency change in the code (1 in main application and 1 in the frequency controller). There should be only one (I have no clue which one is better to keep). 😉