Good afternoon all,
I received a few reports about problems that Argo has on Win XP and/or
2000,
and a few others where similar problems seemed to surface when running on
slow laptops, usually Pentium 166 MHz with Win 95.
After some trials, and thanks to a suggestion of Wolf, I succeeded in
reproducing
the problem on my PC, which, by today standards, is also a slow machine
(an Athlon
750 MHz), but fast enough to prevent me to spot the problem earlier.
What apparently happened was a buffer starvation during high CPU load peaks,
where the buffer pool I had allocated was emptied before the program had
a chance
to requeue them to the sound subsystem, after processing.
I have changed the logic of the program in this section, and now it
seems to work ok,
at least on my PC. This means that I am now unable to reproduce the
problem,
but I can't have an absolute certainty.
This is where I need your cooperation, expecially from those who
reported the
malfunction, no matter whether on Win 95 or XP or 2000.
I have uploaded a test version on
http://www.weaksignals.com/bin/argo130test.zip
Please unzip it in your Argo directory, so that it can use the DLL,
which hasn't changed.
If your reports will be positive, then I will change the released version.
BTW, while at this, I discovered a quirk of the Borland compiler
TRegistry class under XP,
which forced me to not use it, calling instead directly the Windows APIs.
This has the unfortunate consequence that the format of the settings has
changed,
so you will have to save them again, using the same or a different name.
Sorry, but
in this case the culprit is not me.
Thanks,
73 Alberto I2PHD
|
