Hi,

I'm having a bit of trouble with $ arguments in message boxes - I want to be able to send several arguments to a receiver but I can't for the life of me see where I am going wrong - here is a simple example showing the state after I click the send message:

[; test foo bar(

[r test]
|
[set $1 $2(
|
[bar 0(

[r test]
|
[unpack s s]
| |
[foo\ [bar\

so it seems like all the arguments are getting sent ok (as they are all received in the second case), but what is happening to the foo in the first case?

many thanks

Neil

Hmm - I've just discovered if the first argument is a number then it wall works as I would expect so now I am even more confused...

what is happening is that foo is the selector in the message "foo bar" and therefore "bar" is seen as $1, and "foo" is missed entirely. the help file all_about_messages.pd has info on selectors. to fix it you can put a

object in between, or add a "list" selector to the beginning of foo and bar. the reason it works with a float is because when the first item in a message of more than 1 atom is a float, pd automatically converts it to a list. (and therefore passes it with the "list" selector).

It's a pedagogical doozy.

For example, try this:
[; test list foo bar(

Now you'll get identical behavior with the dollarsigns and [unpack]. So let's confuse ourselves even more:

[; test listless foo bar(

Oops, it's different again. What's going on?

It turns out that [unpack] and some other objects have a "convenience" method if you send them something other than a float, list, or symbol message. Message boxes don't do that-- if you look at the Pd window you'll see that the message box gives you an error:

"$2: argument number out of range"

Unfortunately it won't highlight the source of the error, which is the [set $1 $2( message box.

That error is a _good_ thing, btw. If message boxes could talk, they might apologize for making dollarsign variables start at one for numbers and zero for everything else. On top of that they might apologize that $0 doesn't actually expand to the selector of the message. But [unpack] would need to apologize for remaining silent when it treats "foo bar" the same as "list foo bar", and leaving you clueless when a subtle bug arises from treating different data as if it's the same.