Re: [Audacity-devel] [Audacity-nyquist] Patch improving Nyquist input behavior

[Gale A. <ga...@au...>]

| From Al Dimond <bus...@gm...> 
| Fri, 23 Oct 2009 13:33:38 -0600
| Subject: [Audacity-devel] [Audacity-nyquist] Patch improving Nyquist	input behavior
> On Friday 23 October 2009 12:57:24 Vaughan Johnson wrote:
> > Al Dimond wrote:
> > > On Friday 23 October 2009 02:27:50 Gale Andrews wrote:
> > >> Hi Al
> > >>
> > >> On VisualStudio, I got the following compile error after running
> > >>  this patch:
> > >>
> > >> Nyquist.cpp
> > >> ..\..\..\src\effects\nyquist\Nyquist.cpp(1070) : error C2668:
> > >> 'pow'
> > >>
> > >>  : ambiguous call to overloaded function C:\Program
> > >>  : Files\Microsoft
> > >>
> > >>  Visual Studio 9.0\VC\include\math.h(575): could be 'long double
> > >>  pow(long double,int)' C:\Program Files\Microsoft Visual Studio
> > >>  9.0\VC\include\math.h(527): or       'float pow(float,int)'
> > >>  C:\Program Files\Microsoft Visual Studio
> > >>  9.0\VC\include\math.h(489): or       'double pow(double,int)'
> > >>  while trying to match the argument list '(int, int)'
> > >>  ..\..\..\src\effects\nyquist\Nyquist.cpp(1072) : error C2668:
> > >>  'pow' : ambiguous call to overloaded function C:\Program
> > >>  Files\Microsoft Visual Studio 9.0\VC\include\math.h(575): could
> > >> be 'long double pow(long double,int)' C:\Program Files\Microsoft
> > >> Visual Studio 9.0\VC\include\math.h(527): or       'float
> > >> pow(float,int)' C:\Program Files\Microsoft Visual Studio
> > >> 9.0\VC\include\math.h(489): or       'double pow(double,int)'
> > >> while trying to match the argument list '(int, int)'
> > >
> > > Really!?!?!  OK.  On each of those lines the first argument to
> > > pow() is the literal 10.  Does it build if you cast each of them
> > > to a double? Or change them to 10.0... although that's probably
> > > just as ambiguous.
> > 
> > You just need to change them to 10.0 (e.g.) or 10.0f or whatever
> >  you're actually intending. This happens pretty often due to
> >  compile differences.
> > 
> 
> Ah, that must be why I see so much code (in Audacity and elsewhere) 
> using forms like 10.0 all over the place when 10 usually suffices for 
> me.  For anyone looking to build with VC, a new patch that hopefully 
> builds there straightaway is attached (I don't have an installation to 
> test on ATM).


Thanks, Al. I'd already modified: 

 newVal *= pow(10, precision);
 newVal /= pow(10, precision);

in the old patch to read "10.0" in both cases and that did the trick for
me. 

I tested the patch cursorily on Windows XP in High Pass and Low Pass to
make sure that typing values in the boxes is now respected when you 
reopen the effect. 

I also changed the default value in Low Pass to verify that changed 
default was used when launching Audacity afresh. 

I tried Pluck, though I don't think it was affected by this bug.


> makes it easier for users to enter values into the text boxes at 
> higher precision than what's usually displayed (it used to be 
> that they could not touch other sliders without having their 
> custom values rounded to the nearest slider "tick").

That works for me in High Pass and Low Pass. 



Gale