[Grace-core] Some notes on pattern matching on the wiki

[James Noble]

Hi Kim 

> I'm afraid I don't really understand your pattern matching semantics pages.

fair enough.  I don't think I understand them either - or rather, I do understand them
but I am very aware we need actual implementations and iterators before we come 
up with a final (hopefully clean & simple & explicable) design.

>  Is there any chance you can expand them, perhaps by adding examples?

There were examples, but (unfortunately) in section 500 which is right at the end!

I will try to sort this out over the next day or so: perhaps come back on monday?

I now realise there are actually _three_ steps here: first, the matching syntax,
second, rewriting the matching lambda into a method request that tries to do the matching, 
and then, third: implementing that method dispatch via a method protocol.

Unfortunately the verison of the  discussion on that page mixes
all this up.  I will try to refactor - have another look on monday...


>  I get lost starting in section 0 where I am not sure what bindings and pattern.boundvariables are to represent (can you give an example) and then in section 1 where I have no idea what the "pattern" is in "pattern.matchObject(pat)".

both were typos. 

> For the syntax, I'm more partial to the Scala version than newspeak.

I figured you would be!


> By the way, I haven't seen convincing evidence yet that we need first-class patterns or combining operators on patterns.  [Consider this part of my continuing efforts to keep the language simple!]


A crusade I share!  
But one of the things I think I see in the newspeak design - that I didn't understand until talking to Gilad at the last workshop,
is that the closer pattern matching is to otherwise normal expressions, the fewer extra mechanisms you need in the language.

I don't fully understand all the details of that, nor how to make it work in Grace.  
But I'd like to have a try. My goals are:
 - simple straightforward syntax, with as little extra as possible
 - simple translation to non-matching lambdas by introducing match protocol
 - simple matching protocol.

We're not there yet (or at least I'm not).  But it gets better each time around the loop.

more later...

James