00:02 <ptolomy> Anybody know what is on the horizon for the next major version of the compiler? or if there is a next major version in the works?

00:18 <beschmi> next release is 3.09 afaik, but i don't know of any big changes

00:20 <beschmi> but there are some extensions to ocaml that are not integrated in the mainline like acute or ocamlduce

00:45 <twobitsprite> I'm trying to figure out a goo way to break long lines... are there strong conventions in place, or do you generally just break and indent where it seems reasonable?

02:43 <twobitsprite> hmm... it seems like changing the contents of the ref that you are matching on has undefined results.... :P

03:44 <Smerdyakov> twobitsprite, you read that in the manual?

19:39 <mellum> I need a data structure like Set or Map where the size is tracked so it doesn't need O(n) to retrieve. Does that exist anywhere already?

19:43 <twobitsprite> mellum: hashtable?

19:43 <mellum> twobitsprite: no, something functional

19:45 <twobitsprite> why does it have to be functional?

19:47 <mellum> twobitsprite: because that makes programming much easier

19:49 <twobitsprite> does this data structure need to be dynamic? i.e. is it going to be constructed on the fly?

19:50 <mellum> Yes. If I just query it, it wouldn't matter whether it's functional or imperative, would it?

19:51 <twobitsprite> correct...

19:54 <twobitsprite> well... updating memory in place is always going to be faster than the copy/substitute methodology, and the only way to avoid ether of those (that I know of) is through some form of linking mechanism, where one member of the data structure canbe created and "attached" to the rest of the structure, but then you get into linking, and to querry this data strcutre from any given element, you must walk these links... thus giving rise to hi

19:56 <mflux> mellum, I think the size could be tracked by simply each node knowing the size of its siblings, no?

19:56 <mflux> that would also allow queries in the form 'how many values between 7 and 42'

19:57 <twobitsprite> mflux: but you still have O(n) to querry the size of the structure... you have to walk through each simbling to compute the total size...

19:57 <mflux> twobitsprite, no, you build the size to each node while you build/modify/balance the structure?

19:58 <mellum> mflux: absolutely, that would work. I'm just to lazy to code that myself

19:59 <mellum> but it looks like I'll have to.

19:59 <mflux> happy coding ;)

19:59 <mflux> maybe you could take some existing one and hack that feature to it

20:01 <mellum> yeah, I'll look into that... *sigh*

23:44 <lispy`> what is the syntax for cascading match-with expressions?

23:44 <lispy`> i've tried lots of things and i get syntax errors

23:45 <monochrom> You probably want to throw in parentheses.

23:45 <monochrom> Note that "begin ... end" is another way to write "( ... )"

23:46 <lispy`> i'll try the parens

23:49 <lispy`> where should i put them?

23:49 <lispy`> around the entire match with ... sturcture?

23:50 <monochrom> put them to clarify scopes

23:51 <monochrom> e.g., match x with 0 -> (match y with 0 -> ... | 1 -> ...) | 1 -> ...

23:51 <monochrom> without the parentheses the computer would misunderstand.

23:52 <lispy`> okay thanks

23:56 <lispy`> so, you can only pattern match on a single variable at a time, does that make it common to use tuples?

23:59 <monochrom> You can match (x,y) with (0,0) -> ... | (0,1) -> ... too.