Hi Burton,

Did you name it by invoking "table $1_array" or did you go in and change the name using the GUI? If you do the latter, the settings aren't saved (although perhaps they should be...!)

cheers Miller

On Tue, Jun 18, 2002 at 03:53:06AM -0700, Burton Samograd wrote:

Miller (or anyone else with more time to find this bug..ahem...feature :),

I ran into quite a frustrating bug when designing my step sequencer.
I'm currently running pd-0.35-test26 on linux. I was using the table object so i could create local arrays for each sequencer using $1. But it seems that when the patch is saved, it saves the array name in the patch file (ie. if you gave an argument test to the subpatch, the array $1_array would be saved as test_array instead of $1_array).

The problem was that when I tried testing with 2 patches the array wasn't being created for each instance since it had already been created by the patch before it was saved (does that make any sense???) so i would get "test_array multiply defined" messages whenever i would try to write to it.

I then remembered an email i read a while back saying that pd now allows $1's in array names so i tried that and it worked like a charm.
My problem is solved but I thought i would inform you of this behaviour, or if you already knew it existed, maybe you could enlighten me on why it is done that way...

Although I don't know the details of the internal structure of the program compeletely, I would think that when saving the table object should somehow hide the array so the name isn't written to the file, since i assume that on loading the table creates the array if it doesn't already exist...

So there's a couple of my $0.02 on the subject.

burton samograd