Re: [yoshimi-user] New Kid On The Block

  • From: Kristian Amlie <kristian@xxxxxxxxxx>
  • To: yoshimi-user@xxxxxxxxxxxxxxxxxxxxx
  • Date: Fri, 14 Feb 2014 20:15:28 +0100

On 12/02/14 23:18, Will Godfrey wrote:

Hi All.

I've just made my first commit to the master branch so please have a crack at
it and make sure I haven't done anything daft!

First an apology to Lieven Moors, who almost a year ago offered a patch port
from Zyn. for the circle and spike ADsynth waveshapes. I had completely
forgotten about this and subsequently re-invented the wheel!

These shapes are now there!

I've also applied the correction for Phaser panning.

Finally I've fixed two minor compiler warnings.

I want to have a good look around yoshimi now, make sure I understand it (a
little!) and do the boring work of chasing down bugs before trying to move
ahead.

Yey!

There are two bugs I know of and one annoyance.


The first bug is that between V0.060.8.rc2 and 0.060.8 {final} a change crept
in that makes a subtle difference to my Overdrive patch, only really
noticeable
at higher pitches. I suspect it's something in 'filters' but am not sure. I
don't know what other sounds it might affect, but I want to nail it anyway.


The second bug is far more obscure and occurs on all version of yoshimi but
only on certain processor/soundcard combinations (AMD 64bit dual core with
MAudio 2496 or Komplete Audio 6 is definitely OK). I found it by chance and it
affects three of my patches. Yiee, Percussion Organ & Stopped Pipe (from my
companion bank). The bug makes Yiee sort of gargle, Percussion Organ loses its
attack, and Stopped Pipe gains an attack that shouldn't be there.

Sorry for only quoting my patches, but these are the ones I know best
so instantly detect any changes.


The annoyance is the wretched 'cascade' of digits after the decimal point on
detune controls. This does *not* occur on debian lenny, but does appear on a
previously compiled image run on later distros. Clearly then it is some form
of
runtime problem. I've identified that there is a change in version of fltk,
and
suspect that Cal used some form of initialisation that was OK on the older
version (although maybe not 100% correct), but falls foul of the later
versions.

If anyone can shed light on any of these issues I'd be extremely grateful.

I don't really have any information on any of those bugs, just a tip
that I often use. In cases like the first audio bug, where I have *no*
idea which part of the program is causing trouble, but I *do* know a
version where it used to work, I often find that the process of
elimination is helpful. That is, if you compile each version of Yoshimi
in a separate directory, and one by one, you remove filters (by using
"if (0) { ... }" in the code), you might eventually track down the
filter that makes the difference. Boring work, but it has worked for me
in the past.

I'm really stumped on the two other issues.

--
Kristian


Other related posts: