fixed recording of sustained midi notes

[serdnab]

Fixes #1700.
This recovers the behavior done by 4b340f7, and later lost.
When a midinoteff comes from midi, and sustain pedal is pressed, the noteplayhandle pointer don't get nullified, so later, when sustain pedal is released, it can catch that pointer to emit the signal that recording of note is finished.

[PhysSong]

I think using && is better than using many if() clauses. In if (A && B), B isn't evaluated when A is false, maybe you know it already.
So it is safe to write like if (nph != NULL && nph->isReleased()) and that will make the code cleaner.

[tresf]

is safe to write like if (nph != NULL && nph->isReleased())

Or if you believe in such a thing... :) if (nph && nph->isReleased())

[PhysSong]

Should the target branch be master? #1700 is in 1.2.0 milestone.

[zonkmachine]

Should the target branch be master? #1700 is in 1.2.0 milestone.

Yes, or we backport to stable-1.2.

@serdnab I've tested this and it works fine. It can't be tested completely because of #3537.
As commented in the review you can cut down the number of if()'s.
Would it be rude to suggest you to have a look at #3537 too?

[serdnab]

Made the shortening of the if's.

Would it be rude to suggest you to have a look at #3537 too?

I will take a look at it.

[zonkmachine]

I think you can write this into an if else with the else above to simplify it further.
And then lines 324-326 can get one tab each.

[zonkmachine]

I will take a look at it.

Awesome!

I restarted the Travis build. It sometimes will fail for reasons that are not related to the changes made.

[serdnab]

Addressed the remarks.
I realized that the "else" is not necessary, if not isSustainPedalPressed(), m_sustainPedalPressed is already false.

[zonkmachine]

I realized that the "else" is not necessary, if not isSustainPedalPressed(), m_sustainPedalPressed is already false.

So, drop the test and just let there be: else

[serdnab]

I believe you misunderstood me.
The if(isSustainPedalPressed()) is necessary because m_sustainPedalPressed can be true or false, what is not necessary is a possible "else" for the if(isSustainPedalPressed()) clause.

[zonkmachine]

I believe you misunderstood me.

I believe you're right there... :)
Looking at this again.

what is not necessary is a possible "else" for the if(isSustainPedalPressed()) clause.

This is true, but the if( event.controllerValue() > MidiMaxControllerValue/2 ) above it still needs the else. You're testing a state, but this is the event that sets the state. Maybe you can show how you're thinking in code instead?

[PhysSong]

@serdnab Or you can drop else if you put current else if clause first, though current code looks better. I suggest keeping this code.

[serdnab]

I am not explaining well, then you are not understanding.
Anyway, the code I intend is the current one.

[zonkmachine]

The only difference I see is that with dropping the test you have a built in safe guard in case of an unexpected state. When we get a pedal off command when m_sustainPedalPressed is false. This may never happen and/or may not have any side effects at all. I don't know if I'm being picky here. If other developers are cool with this I vote merge.

[PhysSong]

@zonkmachine I suggest merging without more change.

[zonkmachine]

OK. Squashed and merged. Thank you so much Andrés for fixing a long standing bug!

[PhysSong]

@zonkmachine backport?