[Ktechlab-devel] Why the SVN trunk is broken. =(

[Alan Grimes]

I've gotten into the awful habit of submitting before I test. =\

I've tried to power-through the bug but I've spent far more time than I
properly have to find the problem.

I stuck a breakpoint into where the wires were being loaded into pin,
they were all good, unless I missed one... But when "localConnectedPins"
is called, it barfs!! (the wire being pulled is utter nonsense, and
it'll segfault when it tries to derefference its pins...)

This reeks of memory corruption... I've locked down both wire and pin
well enough that they should be very easy to audit, so there's something
bugging up the memory. -- I'm pretty sure this is related to why the
program misbehaved but didn't crash due to over-reliance on
qguardedptr... =\

What might be going on is that something is deleting the wire but not
telling any of the pins...

THAT WAS IT!! I fixed the destructor in Wire, should be good now.

In other news, I assembled all three major sub-units of my DAC, the
clock decoder, the input register, and the output buffer into one single
HUGE circuit and ran it at 1khz, and was most satisfied with the
results. =)  (only one channel/8 bits actually implemented still around
50 gates and flipflops.)

My current brainstorm is a preamp using a single opamp such as the
fairly good TL082 that you can get from some ratshack stores. The
problem with usual preamp is that it squanders its dynamic range by
using an attenuator followed by a fixed (usually large) gain stage. My
solution is to use a quad-gang switched attenuator volume control where
the first two decks are attenuation, and the second two decks are stuck
in the opamp's feedback loop where they can select from 0- to +10db of
amplification. =P They're both wired so that when you turn the control
it smoothly shifts from attenuation + buffer to amplification with not
one db of gain wasted! =P


-- 
New president: Here we go again...
Chemistry.com: A total rip-off.
Powers are not rights.