[PD] abs~ on 64-bit and 0.42.3

[Miller Puckette]

Hi all,

I juect checked and abs~ and exp~ are indeed broken in 0.42-3 (and 0.42-4) -
that got fixed in my source a few months ago but I haven't been releasing
anything since there's a huge GUI re-write underway.

Here's the damage...

[msp at slash src]$ diff d_math.c ~/pd/src/d_math.c 
630c630
<         *out = exp(*in1);
---
>         *out++ = exp(*in1++);
723c723
<         *out = (f >= 0 ? f : -f);
---
>         *out++ = (f >= 0 ? f : -f);

I think I have to try the latest GUI rewrite code (Hans-Christophe and Hannes
are working on it and I'm basically just waiting for it to stabilize) -- and
if it's not at least stable enough for me to put out as a "test" version, I
should probably go back and apply all the applicable bug fixes I can find to
0.42 (a dangerous prospect since that process often seems to _introduce_
bugs as well as fix them!)

cheers
Miller
On Tue, Jan 19, 2010 at 11:34:01AM +0100, IOhannes m zmoelnig wrote:
> Orm Finnendahl wrote:
> > Hi all,
> > 
> > sorry, forgot the attachement. Here it is again...
> > 
> > Hi all,
> > 
> >  abs~ seems to be broken on my machine with 64-bit linux and pd 0.42.3
> > 
> > Clicking in the bang in the attached patch should result in a straight
> > line at +0.5 (at least that's what I think it should do).
> > 
> > Instead the result is a straight line at -0.5 with only two points at
> > arrayindex 0 and 64 set to +0.5
> > 
> > I checked the sources which were used to compile the program and they
> > look ok.
> > 
> > It feels like a 64-bit problem somehow (maybe some compiler flags I
> > didn't set up correctly), but then again I shouldn't rely too much on
> > feelings even though it is a computer.
> 
> there is a bugreport that might be related to this:
> https://sourceforge.net/tracker/index.php?func=detail&aid=2658537&group_id=55736&atid=478070
> 
> 
> mfadr
> IOhannes



> _______________________________________________
> Pd-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list