Closed jackokring closed 8 years ago
LMMS 1.0.3 is an outdated version and its not supported. Please check if this bug exists on 1.1.3, which is the latest stable version. If you expirience the bug on 1.1.3, please offer steps to reproduce and your OS information.
You can get the latest version over at https://lmms.io/download/
Will check when I get the master branch to compile. Currently cmake errors. Debian 8.2 latest apt-get toolchain, forked repo of master.
git checkout stable-1.1 ... cmake .. -DWANT_VST=no make
So I have 1.1.3 still no mallet sound beyond tubular bells
Reminds me slightly of this https://github.com/LMMS/lmms/pull/1529.
Side note.... @jackokring if you wrap your command/codeblocks in:
[stuff]
tags, they'll display nicely. e.g.
echo "foo"
or for c++
[stuff]
OK, minor update, on inspection (listening) of the sound not output, the first time a selection is made of a setting lower than tubular bells, a sound similar to a mangled buffer is played, then nothing ... until one of the working ones is selected. Further more the mangle is heard just once, and never upon reselection of a non working variety.
Reproduced on elementary OS Freya on the master branch.
None of the settings lower than Tubular Bells produce sound, but when I check the 'Bowed' checkbox, they start making a glitched sound, as if they are stuttered ...
I check the box and the sound worked, so seems to me the box is implicitly ticked and should be removed from the GUI.
@jackokring can you please retitle and fix the original description?
So it's the bowed instruments then. It could be the stk package.
Yeah, this is in 0.4.14 too.
@musikBear You reported on this issue on the mailing list around the 1.0.3 release. What version of stk do you have?
He'll have whatever was bundled at the time of 1.1.3
/mingw32
OK. From the release notes for stk 4.4.3
Updates to Bowed class (thanks to Esteban Maestre!)
This places us in time around Ubuntu 12.04 and lmms-0.4.10
lmms-0.4.2 / stk-4.2.0 Same issue, so the changes in stk above had nothing to do with it.
@zonkmachine / current Systems Tool Kit c++ is the one @tresf bundled in the test RC g23d2824. This RC is '100%' stable for me. The new edit-features in piano-roll are incredible useful (saying that not because i suggested those :p, but current stk is ofcause of no relevance to the one at 1.1.3 release time. If i Can make any kind of experiment, that can be of interests? just let me know
If i Can make any kind of experiment, that can be of interests? just let me know
Yes. Just test mallets and see if you find anything of interest.
Yes. Just test mallets and see if you find anything of interest.
@zonkmachine Made a testbed on tresf new x32 (1.1.90 gd098a39) with escalating automation for all instrument-dials. I used wave-recording to illustrate the various functionalities. The first series are for the presets: Marimba, vibraphone, agogo, wood1, reso, wood2, beats, two-fixed, clump That all have the same controller-dials Hardness
The dial does not change anything. doubt it connected.
Position
The dial does not change anything. doubt it connected.
Stick mix
The dial does not change anything. doubt it connected.
Vib Gain The dial does not change anything. doubt it connected.
Vib freq Looks fine
Spread Waveform not influenced, but it is ok for *all.
Second series Preset Tubular bells Modulation
Crossfade
ADSR Looks fine
LFO speed Looks fine
Third series Presets Uniform-bar, Tuned-bar, glass, Tibetan-bowl These 4 presets is not working at all. They produce no sound unless bowed, but it is only 'noise'. In the case of Uniform-bar extremely loud, and dangerous for ears/speakers. ! Be careful !
@musikBear We got pretty much the same result then. I'll add to this that:
:+1: very good
@LMMS/developers What I'm thinking about this right now is:
This way you end up with two new instruments that works and has more manageable code.
Edit: An updated Mallets will open older projects wrong because people will have turned the knobs. Even though that had no effect previously it really saved the settings so it will come out wrong.
An updated Mallets will open older projects wrong because people will have turned the knobs
Maybe that could be addressed using DataFile::upgrade()
?
Maybe that could be addressed using DataFile::upgrade()?
^What Fastigium said, making another version of the same plugin is the wrong approach to this problem imo.
Maybe that could be addressed using DataFile::upgrade()?
I'm sure there's way to manage the update but we need to take care of the other bugs first and they're in code related to/or in, the BandedWG module. https://github.com/LMMS/lmms/pull/2671#issuecomment-195940322 If you find what's wrong in there bump me about it but for now I'm stuck. I just think the most clever way is to roll along with the two spin-offs of Mallets and instead find a way to hide old stuff in the menu and only launch them in project's that already got them and for users who choose to activate all plug ins.
If you find what's wrong in there bump me about it but for now I'm stuck
My plate is full with crash-fixing for the moment, but perhaps you might gain some insight by using conditional breakpoints in gdb. They should allow you to halt execution when a spike happens and trace where it might come from (bt full
will show you values of local variables). Hope that helps, otherwise I'll see if I can take a peek sometime soon.
OK. I'll try and move my upgrades to Mallets instead and will read up a bit on gdb.
You may also find gdb's frame
command useful. It allows you to enter the scope of any function in the backtrace and print members of local objects and such. Was invaluable to me in fixing those "index out of range" crashes.
Mallets is now ready for testing. https://github.com/LMMS/lmms/pull/2670
I used the same backward compatibility method as Kicker as I recently worked with it and it's familiar. Most bugs fixed. Mallets seem more to have been abandoned before really completed and left with more or less the preset values from stk modalbar, BandedWG presets turned default off and only Tubular Bells from stk::FM really working full out.
They should allow you to halt execution when a spike happens
I haven't been able to reproduce the spikes.
The first time a selection is made of a setting lower than tubular bells, a sound similar to a mangled buffer is played, then nothing ... until one of the working ones is selected. Further more the mangle is heard just once, and never upon reselection of a non working variety.
Edited @tresf