Sending info to an array via message
Hi,
I notice that when sending a list of values to an array via a message, you need to add an extra value at the beginning of the list. For ex, if I want an array named "table" to have the following values {1, 1, 0, 1}, I need to send this message
[; table 0 1 1 0 1(
Am I right? any explanation why? I feel I miss something here, maybe the syntax/architecture of a message...
thanks
ymotion
for example,
if you send {5000, 1, 1. 0. 1}, then you will start writing those messages at index 5000.
in practice, i have never once written a table like that, but that's how miller programmed it, and to change it now would break backwards compatibility, so we're stuck with the slightly confusing index value needing to be attached to the start of every list send to a table.
maybe some other people use that function? i doubt it...
Hi Mod,
Thanks for clarifying. It was indeed confusing...
So, is there another easy way to write values into an array? I mean, other than using [tabwrite] which I sometimes use for functions, but for sending a list of say 10 values, it seems a bit cumbersome process.
Thanks!!!
ymotion
It gives me a great idea for using a single table where I have been using many tables to record n dimensional signals.
Rather than using N independent arrays, I can stripe the data into a single array. This is much more compact to code, and write and read whole entries using (index*N) like so:
#in this case, N=5
[index(
[fivedimensionarray $1 $2 $3 $4 $5 $6(
thanks!
any technology distinguishable from magic
is insufficiently advanced.
Hi Sunji,
Thanks for your reply. Could you write a short patch and attach it. I'm having difficulties to understand just by text.
the syntax with the colon is a shortcut for send. so to send lists quicker:
suppose you have array myarray:
[4 7 2 27 7( <-- some list (could come from whatever list processing)
|
<--instead of 0 put whatever index you want to start writing at
|
[send myarray]
sunji: you might be interested in how multi dimensional arrays are handled in memory, usually they are arrays of arrays stored in blocks directly next to each other (in 1 array) which is what you've discovered in pd. for instance to get row 2, column 4 of a two dimensional array you would take the address of the beginning of the entire array and add to it (the pointer): ((size of columns * desired row) + desired column).
internally the compiler sees the column array as a data structure of it's own, and makes an array of that data structure
nitpicking: technically what you describe is not a five dimensional array but a 2 dimensional array with one of the arrays being of size 5
any technology distinguishable from magic
is insufficiently advanced.
Hi Sebfumaster,
Got it! Thanks for clarifying
Still curious though about Sunji's [index( message. What did he mean by that, and which object exactly should receive the message?
Just got Sunji's reply...so I'm looking into it now
Hi Sunji,
Ok, very nice! I like your idea of writing several arrays into one.
Now I'm curious: isn't there a way to write a matrix in Pd? Lets say you have 5 arrays, you can then have one matrix with 5 rows, no? But, again, I know nothing of matrices in Pd...
Interessing. I tried to blow it up to 3D.
Oops! Looks like something went wrong!
