Filippo Gentile

I'm getting new Udp pachet drops. I'm starting to suspect this is related to `Z21ServerEmulator` doibg everithing in GUI thread. Which means ma y packet can accumulate before it's ready...

Issues were on both Traintastic and ServerEmulator sides. They are now fixed by recent commits plus PR #73

Seems be fixed by running directly `sudo cmake --build . -t install`

Like if install steps which require root privileges are mixed with normal build steps. If later I normally build again I get this: ``` Could not open:~/Documents/winft/como/build/plugins/effects/private/meta_types/effectsplugin_json_file_list.txt.timestamp gmake[2]: *** [plugins/effects/private/CMakeFiles/effectsplugin_automoc_json_extraction.dir/build.make:71:...

> Hmm, that's a weird error. Didn't you have similar issues with LXQt build? At least I think I remember that it says something about moc and timestamps there too....

> About the Z21 kernel change, what issue do you run into? > > If `setValue` is called with the same value the property already is, it won't fire a...

> I think we should fix this in the `DecoderFunction`, when the value gets updated by all the kernels it should do the timeout magic and use `setValueInternal` like before,...

> I fact we should use this _"pattern"_ for all such hardware values. Then we could open a draft PR and see where this pattern lead us to. Maybe it's...

@reinder Hi, I've rebased on current master and tried your `updateValue()` suggestion. I like the principle but the problem is still relevant. Interfaces call `Decoder::setFunctionValue()` which internally now calls `DecoderFunction::updateValue()`...

I've quickly read #178 Well, we are not that far from it. Now interfaces use `updateDirection/Throttle/EStop/Function()` methods. And rest of the code should use direction.setValue(). But it's not explicitly used...