IRC log started Fri Oct 8 00:00:00 1999 [msg(modtunes)] permlog 1999.1008.mod -:- hcf [nef@me-portland-us109.javanet.com] has joined #modtunes -:- mode/#modtunes [+o hcf] by ChanServ -:- SignOff hcf: #modtunes (Leaving) -:- hcf [nef@me-portland-us914.javanet.com] has joined #modtunes -:- mode/#modtunes [+o hcf] by ChanServ * hcf/#modtunes is away: (afk) -:- SignOff hcf: #modtunes (Ping timeout for hcf[me-portland-us914.javanet.com]) -:- hcf [nef@me-portland-us644.javanet.com] has joined #modtunes -:- mode/#modtunes [+o hcf] by ChanServ -:- SignOff hcf: #modtunes (Leaving) -:- water [water@ppp-ip145-c16.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ -:- Beholder [beholder@ppp-069.m2-2.sub.ican.net] has joined #modtunes -:- mode/#modtunes [+o Beholder] by ChanServ -:- Tril [dem@bespin.dhs.org] has joined #modtunes -:- mode/#modtunes [+o Tril] by ChanServ Whoah... I got an invite ;) sigh... this is what i have put up with since i returned 14-year-olds water: Ahh, it's always like this, IRC is free, etc... lots of them i need to figure out how to describe my evaluation model it's like water pipes hehe heh or electrical circuits you create "continuations" by calling partial evaluate until there are no arguments left then you hook the continuations up the the water supply and then they become active information flows through them hmm 05:40pm any comment? sounds exactly like a category well, sort of how is there an evaluation model in CT? well, arrows model lambdas, and nodes types how do you differentiate an expression that is actively being evaluated (arrows followed, values at the nodes changed in response to ones upstream), from one that is dead, not evaluating? the picture resembles your pipes idea, because you can sequentially-combine lambdas that way It's not even part of the picture so, your pipes are not functions? they're *instances* of functions, *with all arguments supplied* oh closures right well, category theory is basically the typed \-calc partial-eval and all still thinkin 05:50pm well, your pipe idea gives lambdas a state, it seems on/off which reminds me of a haskell paper describing a persistent store for the language as storing graphs of closures whadya think? water:holdon a sec k back k first i'm going to explain the persistent store assume it is a collection of cells (variables) each cell has a type, and it varies over the possible values of that type 06:00pm (type=set of values) yes... there are then compound cell types such as tuples which contain more than one cell .. one set of values that could bea type is the set of all cells so you can have cells that reference other cells. tuples: which can be constructed via the usual lambda constructions well, tuples are constructed by providing a function that given the tuple, returns an individual cell in it (one function for each element in the tuple) brb.. i will get my CT book Anyway, using reference cells (cells that refer to other cells), you can construct expressions... b sure the idea of information flow- you can connect a cell to the output of a calculation and you can also connect cells to the input of a calculation so when the inputs change, the output is automatically updated data-flow style mmm.... sounds like a whole lot of arrows being thrown around :) I say this is allowed by using partial evaluation where a CELL is the argument, not just a static value. if I P.e. a function that has a cell as the argument, the result is a function that always reads from the cell for that argument (that's how the argument disappears) 06:10pm which to me just adds a layer of referencing to the actual data but to maintain the illusion of the function being different for different values of the argument, it makes references implicit grok'd the function has to watch for changes in the argument like spreadsheet agents? sure. yeah, that's like data-flow ideas well, it sounds good so far I want to have the entire system be a giant data-flow diagram, and the "dynamic compiler"'s job is to "implement" the data flow on the fly and thus implement the system (my personal choice is to leave the referencing explicit in general as an ontology system) yeah, i was thinking about that for a while... i never really left the idea, but i did forget the perspective implicit referencing to Persistent Store cells, is a clean way to formalize side-effects say, that _is_ an interesting take on my problem for making the arrow code 'dynamic' sure, but the persistent store idea then must be consistent brb also, a key part of my system is the backwards-partial evaluation 06:20pm -:- SignOff water: #modtunes (Ping timeout for water[ppp-ip145-c16.tscnet.net]) hey, when did water leave. I didnt see a leave message 21:27:38 on my machine sorry I was afk -:- SignOff Beholder: #modtunes (Read error to Beholder[ppp-069.m2-2.sub.ican.net]: EOF from client) 06:40pm -:- water [water@ppp-ip145-c16.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ -:- water [water@ppp-ip145-c16.tscnet.net] has left #modtunes [] -:- water [water@ppp-ip145-c16.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ water, you here? finally i am way too nice to my roommates using my account to read up-to-date news from wcw and wwf is _not_ my idea of a good use of bandwidth 06:50pm where were we? backwards partial-eval fits well in CT, btw factorization of types... arrows can model morphisms which can encode the embedding of types into tuples reversability in that case means reversal of the idea Tril: listening? yes... how does that tie in with type products? are you using tuples as the result of type products? yeah there is a distinction between, persistent store entities that combine two cells, and one persistent store entity that contains a product type it's tricky.. because you can go from one to the other yes it is and usually you can't tell which one it really is but it probably doesn't matter :) define "combine two cells", though 07:00pm er.. combining two cells doesn't have to be the same thing as a strict product it would be a cell that was a product of two reference-to-string-cell types... instead of a cell that was a product of two strings -:- SignOff water: #modtunes (Ping timeout for water[ppp-ip145-c16.tscnet.net]) 07:10pm -:- water [water@tnt-9-136.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ -:- SignOff water: #modtunes (Ping timeout for water[tnt-9-136.tscnet.net]) -:- water [water@ppp-ip84-b19.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ i was thinking... what about the idea that reflection is just a subset of the actions that a program can take? er... not well expressed 07:30pm So what we're doing is designing a language that makes source code easy to be programmatically modified... current languages don't make that easy. -:- SignOff water: #modtunes (Read error to water[ppp-ip84-b19.tscnet.net]: No route to host) -:- water [water@ppp-ip84-b19.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ wb... :( as you were saying? hmm 07:40pm i lost my train of thought reflection is a subset of actions a program can take? But using reflection you could express ALL the actions a program could take.. well, i thought i mentioned that this is what makes re-write logic so attractive for reflection but it didn't make it to the channel (yes, it took that long for me to get the logs) water: I'm having problem distinguishing some slightly different notions of type 1. a set of values 2. abstract types, which are defined by what operations are available on them (and the set is left open ended- anything with those operations is a member, and thus the "values" of the type are not all immediately available) hmm water: did you reply about what I said here, in the other channel? about confusing? er.. not intensionally. it applied to both it was supposed to be for the other discussion, but applies here too anyway... you can often define one from the other in the #2, I can have a "non-hierarchy", where the top (there is a top, despite that it is not a hierarchy) has 0 functions applicable to it and the next layer has 1 function, and the next has 2 (making each level geometrically bigger, hence the non-hierarchy) in #1 it's quite clear-cut and the set of operations is separate from the values themselves define one from the other? 07:50pm -:- SignOff water: #modtunes (Ping timeout for water[ppp-ip84-b19.tscnet.net]) -:- water [water@ppp-ip110-b45.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ :P ok when the source code is "off" ie. not being evaluated, that means it is not currently the argument to an evaluating interpreter. An evaluating program is likewise simply some source code that is a dependent of a running interpreter follow? -:- SignOff water: #modtunes (Read error to water[ppp-ip110-b45.tscnet.net]: No route to host) -:- water [water@ppp-ip110-b45.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ 08:00pm <_water> tril: you can use an adt to define a set, as in peano's axioms -> natural numbers <_water> tril: a set can also define an adt i said it earlier, but was cutoff before i descovered that the logs didn't have it using the multiple-counting numbers what is it claled,? Omega? anyway, multiple dimensions of counting, I guess i'll take your word for it, if youcan define a set (ie. a unique identifier for each element) then it should be implementable with an unbounded-value representation how do you make adt from a set? -:- SignOff water: #modtunes (Ping timeout for water[ppp-ip110-b45.tscnet.net]) 08:10pm -:- water [water@tnt-9-11.tscnet.net] has joined #modtunes -:- mode/#modtunes [+o water] by ChanServ -:- SignOff water: #modtunes (Ping timeout for water[tnt-9-11.tscnet.net]) helo are you back 09:00pm * Tril/#modtunes is away: (afk) [BX-MsgLog Off] [msg(modtunes)] newlog 1999.1009.mod IRC log ended Sat Oct 9 00:00:01 1999