Make SampleBuffer adjust its members when resampling. Fixes #5218 for stable-1.2

[softrabbit]

My proposition on how to make oversampling work again in the 1.2 line after the changes in #4991. Feel free to chime in if some use case of SampleBuffer remains broken (or gets a fresh bug) after this PR.

[softrabbit]

Oops. This won't work, just realized I must've only tested the latest changes from command line. The loop points will be off when returning to normal sample rate after GUI export. I'll yet be refining this.

[Reflexe]

I have heavily refactored this class in the recording branch. Maybe it would be smarter to use that code instead (Will do that hopefully later this day).

[softrabbit]

I have heavily refactored this class in the recording branch. Maybe it would be smarter to use that code instead (Will do that hopefully later this day).

In master, no doubt. It's meant to end up there anyway.

In stable, why not if you already fixed the bug in your branch. If not, I'd be inclined to not backport and just fix. But either way works for me, you are most familiar with your code so you get to decide.

[Reflexe]

Looks good, however, maybe it would be better to just avoid normalization altogether? (let play handle resampling).
That will also make sure we won't have to resample again after we finish exporting (which will may reduce quality)

[softrabbit]

Looks good, however, maybe it would be better to just avoid normalization altogether? (let play handle resampling).
That will also make sure we won't have to resample again after we finish exporting (which will may reduce quality)

I believe update() calls normalizeSampleRate() only after reloading the original data from file, so there shouldn't be a successive resampling problem.

[Reflexe]

believe update() calls normalizeSampleRate() only after reloading the original data from file, so there shouldn't be a successive resampling problem.

You're right. I just hate this update function so much ;)

I think it just would be better to extract (float)mixerSampleRate() / old_rate into one variable and we're done.

[PhysSong]

Is it okay if I(or someone else) add a commit addressing #5226 (comment) and merge this?

[softrabbit]

Is it okay if I(or someone else) add a commit addressing #5226 (comment) and merge this?

Sure. I will probably not be much around my dev box for the next few days, so go ahead.

[PhysSong]

Okay, but I'm not sure how to name the variable.

[zonkmachine]

It looks like this is still an issue on drumsynth (.ds) files.
https://lmms.io/forum/viewtopic.php?f=7&t=31122