Hi,
I'm struggling with a problem of the most weird kind.
I've not been able to isolate it, but I do can report a systematic behaviour which astonishes me.
PD-Vanilla 0.40.1 in Windows XP
I have an enormous very complex patch that I've been developing for months, built with a careful top-down/bottom-up design (i.e. designing and reusing a huge lot of abstractions at various levels), with a lot of GOP, also nested. Up to now I didn't have any serious problem, and the programming of the patch is quite clean; for example I used to be able to use it for hours without without a single error message ever showing up in the output.
Now, after some recent changes, the ABSURD behaviour is:
*CASE 1* -I launch PD -I open my main patch -PD crashes immediately
*CASE 2* -I launch PD -I create a new patch (File/new) -I put a [route a b c] (literally) object, and NOTHING more -I open my main patch -PD DOESN'T crash and the patch works great as it used to do.
If I launch PD with from commandline with -noloadbang, it never crashes, so I guess the crash is caused by some message triggered by some loadbang
I've repeated the two tests tens of times (always closing pd and starting over from scratch) and the results are 100% coherent, i.e. case 1 always crashes, and case 2 never crashes. I'm really astonished: how is it possible that the mere existence or non existence of a [route] object not connected to anything can prevent or not the crash, whatever causes it?!!!???!?!?
I know I'm giving too little information to ask for a diagnosis, but does anybody know of some weired bug of [route]??
The "recent changes" mentioned above, are in the [control_element] abstractionn that I attach, which uses the [schange] and [switch2] abstractions. I can't see anything wrong in them.... I can't reproduce the bug with a simple test patch using an instance of [control_element]. It is used tons of times in my "main" patch.....
Any idea? Do you see any potential problem in the attached patch that may cause illegal situations under some circumstances?
I will greatly appreciate any help
Bye m.
matteo sisti sette wrote:
I'm really astonished: how is it possible that the mere existence or non existence of a [route] object not connected to anything can prevent or not the crash, whatever causes it?!!!???!?!?
most likely you are experiencing memory corruption. the instantation of another object ([route]) will just change the memory layout sufficiently to not expose the bug.
I know I'm giving too little information to ask for a diagnosis, but does anybody know of some weired bug of [route]??
as said above, it is unlikely that [route] is directly related to your trouble. it just helps to (not) expose the real problem.
this kind of bugs is generally hard to track down. i would suggest to create a list of all objects you are using, do a ranking of the likelihood that they are the culprits and investigate them separately.
also make sure that you are using the latest and greatest (well, the greatest might be sufficient) release of any externals you are using.
fmga.dsr IOhannes