01:16:26 billh left #tunes
02:38:15 smoke joined #tunes
03:31:04 jut joined #tunes
03:32:46 [QUIT] jut quit: [x]chat
03:38:54 [QUIT] billh quit: BitchX: use it, it makes hair grow in funny places!
03:55:48 eihrul joined #tunes
05:55:19 [QUIT] air quit: http://www.qzx.com/ :: 
06:21:38 Ghyll joined #tunes
06:40:10 <eihrul> hey
06:41:54 <Ghyll> hi
08:08:52 Nick2 joined #tunes
09:17:11 [QUIT] Nick2 quit: Ping timeout for Nick2[usr41.atil.net]
09:45:42 smklsmkl joined #tunes
09:45:54 [QUIT] smoke quit: Ping timeout for smoke[10dyn154.delft.casema.net]
09:47:24 smoke joined #tunes
10:13:48 [QUIT] Tril quit: [BX] Ping-pong timeout
11:05:25 [QUIT] eihrul quit: [x]chat
11:29:59 lar1 joined #tunes
11:54:26 water joined #tunes
11:54:35 <water> hi all
11:58:48 <water> does anyone have something to discuss?
12:07:31 <Ghyll> water: do you suffer from familiary with postgresql ? 
12:07:42 <water> heh
12:07:51 <water> no, but i do use sql quite a bit
12:08:31 <Ghyll> in postgresql, you can add new basic types to the sql language
12:08:40 <Ghyll> say, int64, or float128
12:08:46 water shrugs
12:08:49 <Ghyll> and you can add functions and operators for them.
12:08:54 <water> ah
12:09:11 <Ghyll> in fact, postgresql is very much like a language environment unto its own.
12:09:19 <water> that's nice
12:09:29 <Ghyll> execpt you can't (yet) add these things using a built-in language.
12:09:35 <Ghyll> it has to be compiled C code.
12:09:35 <water> hm
12:09:39 <water> eww
12:09:49 <water> hmmm
12:09:57 <Ghyll> I've been thinking of bridging the last river, there, just for the heck of it.
12:10:09 water contemplates if smalltalk could be bound to it
12:10:14 <Ghyll> then you would have a lisp-alike database. 
12:10:21 <Ghyll> most certainly.
12:10:35 <water> squeak already has a mySQL binding
12:10:39 <Ghyll> there are bindings for python, perl, c, c++, etc, so integrating smalltalk into it should be possible.
12:11:07 <Ghyll> of course, doing what I want would just be a silly hack.
12:11:38 <Ghyll> tho it'd be nice to have a environment which was built on a real database, with replication and failover and all..
12:11:49 <water> hm
12:11:52 <water> i dunno
12:11:58 <Ghyll> could be an enviroment where all programmers worked together simultaneously.
12:12:05 <Ghyll> in sort of a hierarchy. 
12:12:12 <Ghyll> passing live objects to each other.
12:12:12 <water> i'd rather have database niceties as a special case of a more general framework
12:12:23 <Ghyll> sure. 
12:12:29 <Ghyll> that's the ultimate goal.
12:12:34 <Ghyll> or, rather, it would be.
12:12:35 <water> and hierarchies of users SUCK :)
12:12:42 <Ghyll> no. not hierarchies of users.
12:12:46 <Ghyll> I should've said namespaces.
12:12:49 <water> programmer=user
12:12:54 <water> same thing
12:13:01 <Ghyll> each user has his own namespace, to send things in and out of.
12:13:01 <water> even with namespace
12:13:22 <Ghyll> how would you avoid name clashes if you didn't impose a hierarchy ? 
12:13:49 <water> how would name clashes arise? :)
12:14:20 <Ghyll> two people binding different objects to the same variable.
12:14:27 <water> so?
12:14:39 <Ghyll> and both wanting to use what they boud.
12:14:41 <Ghyll> bound
12:14:41 <water> if they have separate namespaces, there's no clash afaict
12:14:50 <Ghyll> exactly.
12:15:03 <Ghyll> oh. you're not against namespaces, just a hierarchy of them ?
12:15:03 <water> but hierarchy is something different!!
12:15:10 <water> yep :)
12:15:14 <water> (see slate)
12:15:38 <Ghyll> I don't have a definite opinion on that subject yet. 
12:15:47 <water> ok
12:16:09 <Ghyll> then a few q's about slate..
12:16:14 <Ghyll> 1) can you have inner classes ? 
12:16:18 <water> i've been developing my reasons for it ever since '95 when i first tried to make that tunes hll thing
12:16:27 <water> classes? :)
12:16:43 <water> classes are easy
12:17:02 <water> but not necessary to the language
12:17:10 <water> and yes you can "nest" anything
12:17:36 <water> slate is just based on namespace graph rewrite
12:17:37 [QUIT] ink quit: bye all
12:18:06 <Ghyll> then you do have something akin to hierarchical scoping
12:18:12 <water> nope :)
12:18:42 <water> because access to "outer" things is just another link ( called a slot)
12:19:05 <water> the hierarchy-like form we have been working with initially is just provisional
12:19:08 <Ghyll> when you in the "outer" space, you cannot see the "inner" slots ?
12:19:24 <water> "see"?
12:19:26 <water> nope
12:19:45 <water> "foo bar baz" is a trace through three namespaces
12:20:23 <water> hm
12:21:46 <Ghyll> I see I've been away too long. Slate has actually been worked on... :)
12:22:02 <water> yes slate needs a drastic update to its docs
12:22:15 <water> btw eihrul's putting together a serious implementation
12:22:46 <water> first step is an interpreter for slate called pidgin (because it borrows from the self implementation)
12:23:22 <water> he's also been working on dynamic compilation/optimization stuff... but that will take longer
12:24:18 <water> i'm doing more research into how to build really novel object systems
12:24:31 <water> like the kind that tunes talks about
12:24:41 <water> hm s/tunes/Fare/
12:25:31 <Ghyll> inspiring
12:25:39 <water> heh
12:25:42 <water> slow....
12:25:50 <water> my job is taking way too much time lately
12:26:18 <water> you know Fare's going for VC now?
12:27:33 <Ghyll> VC as in venture capital ? 
12:27:36 <water> yep
12:27:40 <Ghyll> wow.
12:27:55 <Ghyll> and tunes will get an abominable license  ?
12:28:00 <water> actually they approached him first
12:28:07 <water> who knows
12:28:18 <water> apparently he's still working out the business model
12:28:37 <water> (yesterday in the logs has a discussion on it)
12:28:38 <Ghyll> has he dropped any hints on what he's planned ?
12:28:42 <Ghyll> oh. ok.
12:28:48 <water> (i wasn't there... 40 hours at work :P)
12:30:19 <Ghyll> wtf ?
12:30:24 <Ghyll> what's up at work ?
12:30:26 <Ghyll> crunch time ?
12:30:32 <Ghyll> war ?
12:30:34 <water> heh
12:30:42 <water> we have to get the reactors re-certified
12:31:06 <water> and an important group of technicians really royally screwed up
12:31:45 <water> (not mine, though ;)
12:31:51 <Ghyll> so you had leakage all over the place ? 
12:31:56 <water> no no no
12:32:16 <water> this group is the one to do the reactor startup paperwork and tests
12:32:41 <Ghyll> oh. a lot of paperwork mess, then ?
12:32:45 <water> they kept making stupid tiny mistakes
12:32:47 <water> sort of
12:33:02 <water> like taking a switch to the wrong position
12:33:16 <water> or writing info in the wrong block
12:33:29 <water> it's really a long story
12:33:44 <water> (which we were told again and again :P)
12:33:55 <Ghyll> no wonder it took 40 hours :))
12:34:11 [QUIT] lar1 quit: yay
12:34:14 <water> and that's just the start of it
12:34:49 <water> those technicians have to do two checks per reactor per day for the next week and then get re-certified
12:35:01 <water> and each check takes ~10 hours
12:35:17 <water> this delays a lot of work (but not all, thank heavens)
12:35:53 <Ghyll> sounds like a heavy chore
12:36:02 <water> yeah, but enough about work
12:36:24 <Ghyll> heh.
12:36:34 <water> seriously it depresses me
12:36:36 <Ghyll> yes. I'm reading up on the slate papers again, looking for things that have change.
12:36:37 <Ghyll> d
12:36:49 <water> i haven't updated the site in months
12:36:53 <water> however
12:37:06 <water> there are a few new things at http://diktuon.arrow.cx
12:37:24 <water> hm if it's up at all
12:37:39 <Ghyll> it woke up here
12:37:54 <water> yeah same here
12:40:22 <Ghyll> however, there's a lot of unconnected nodes around..
12:40:27 <Ghyll> 's/'re
12:40:34 <water> yep
12:40:40 <water> diktuon was down for a long time
12:41:02 <water> it interrupted the writing process quite a bit
12:42:41 <Ghyll> what kept you from using wikiwikiweb ?
12:43:01 water shrugs
12:43:06 <water> too difficult to organize
12:43:49 <water> i have enough trouble just trying to outline the ideas in an appropriate manner
12:52:38 <Ghyll> there's some new stuff there I've not seen before.b
12:52:51 <Ghyll> but it's all a bit of a haze for me.
12:52:57 <water> yeah have you seen the updated old tutorial?
12:53:18 <water> the syntax no longer relies on lisp's parentheses
12:54:34 <water> none of this matters until i manage to get the full picture and document it properly
12:54:53 <water> eihrul and i have been developing slate as part of our (private) discussions
12:55:35 <Ghyll> aha.
12:56:11 <Ghyll> I didn't see anything in the tutorial change on your home page
12:57:14 <Ghyll> ah, yes. there's some changes.
12:59:30 <water> the "."s don't all act the same
12:59:46 <water> since they are just the apply phases of the meta-objects
13:01:12 ult joined #tunes
13:01:21 <water> re ult
13:01:24 <ult> 'lo
13:01:56 <Ghyll> water: what are the implications of storing the state in the 'parent' slot ? 
13:02:38 <water> it turns the object into a data-flow value
13:02:55 <Ghyll> could you elaborate ?
13:03:02 <water> hm
13:03:45 <water> the meta-object figures into this
13:03:56 <Ghyll> why am I not surprised.
13:04:16 <water> but basically the idea is that if "x" is supposed to have a state of 4....
13:04:18 <water> P
13:04:48 <water> then it makes no difference whether when asking for "x" one gets 4 or a reference to 4
13:05:06 <water> now, when trying to mute x's value....
13:05:33 <water> that's where the mo comes in, because you have to change the namespace structure
13:05:57 <water> *how* that structure gets changed is what the MO's apply phase determines
13:06:28 <water> basically, making "parent" work to emulate state is just simpler for meta-programming
13:07:01 <Ghyll> I see what you're saying but I can't figure out the (implementation) details yet.
13:07:05 <water> and it's one less syntactic primitive
13:07:10 <water> hm
13:07:43 <Ghyll> reducing the syntactic primitives is generally a Good Thing. 
13:07:54 <Ghyll> but does this clutter up the implementation noticably ?
13:08:06 <water> not necessarily...
13:08:17 <water> for example, we're starting from the self implementation ideas...
13:08:34 <Ghyll> from the ideas, or from their implementation ? 
13:08:35 <water> where an object is just a map from slotnames to values
13:08:55 <water> the implementations
13:09:13 <Ghyll> so the first implementation(s) of slate will be based on the self code ? 
13:09:19 <water> in our case, we don't have to treat assignment as anything other than a case of general slot access
13:09:47 <water> not really, eihrul's making it from scratch
13:09:54 <Ghyll> ah, ok. now I follow you.
13:09:59 <water> there's a lot of problems with the self VM code
13:10:17 <Ghyll> is slate supposed to run in a VM ?
13:10:22 <water> no
13:10:45 <Ghyll> on that note, how do arrays figure into this ? 
13:10:52 <water> we want to self-host the compiler whenever possible
13:10:57 <Ghyll> they would much like they do in lisp ? 
13:11:04 <Ghyll> much/work much
13:11:04 <water> heh glad you asked :)
13:11:37 <water> arrays have special hw types in pidgin, but at the language level, it's transparent (relatively)
13:12:10 <water> basically you have an object whose meta-object allows you to use all of the integers as indexes
13:12:17 <water> er...
13:12:25 <water> the slot names are integers, rather
13:13:22 <Ghyll> does that preserve the semantic notion that there are 5 slots beteen 0 and 7 ? 
13:13:51 <water> 5?
13:14:02 <Ghyll> 6.
13:14:04 <Ghyll> my mistake.
13:14:05 <water> don't you mean 6?
13:14:08 <water> oh :)
13:14:18 <water> it depends
13:14:29 <Ghyll> (I've been without medication for a few days, so I'm a bit groggy..)
13:14:33 <water> the idea is that slot-lookup can be modified
13:14:49 <water> so you can have either semantics
13:15:12 <Ghyll> how do you image array lookups will perform in practice ? 
13:15:15 <water> well, any semantics, really, but only a few will be primitive in pidgin
13:15:36 <water> hm
13:15:49 <Ghyll> comparable to java or c ?
13:15:54 <water> every object, when activated, will respond according to its MO
13:16:02 <water> oh speed
13:16:07 <water> i don't know
13:16:16 <water> probably relatively slow at first
13:16:43 <water> like bytecode vm speed for the first round
13:17:35 <Ghyll> that doesn't tell me much :)
13:17:45 <water> of course it doesn't
13:17:49 <Ghyll> if you have to do a named lookup for every access, it'll be _slow_.
13:17:56 <water> that's why i said " i don't know"
13:17:59 <water> geez
13:18:13 <Ghyll> yesyes. I'm on your team, remember ? :)
13:18:17 <water> no, named access is not literal
13:18:44 <water> it will get compiled into some IL that eihrul is working on
13:19:19 <Ghyll> sounds promising
13:19:50 <water> there's a lot of work and writing to do yet
13:19:54 <water> and thinking
13:20:04 <Ghyll> in how much detail can a meta-object go wrt the object layout in memory ? 
13:20:19 <water> potentially, unlimited
13:20:33 <water> all it has to do is support the object model
13:20:58 <Ghyll> will the meta-object ever be force to supply low-level IL code for its semantics ?
13:21:05 <Ghyll> force/forced
13:21:10 <water> hm
13:21:30 <water> it depends on what stage of the implementation we are at
13:21:41 <Ghyll> how do you mean ? 
13:21:55 <water> currently, all MOs if not primitive have to be compiled into IL
13:22:13 <Ghyll> let's imagine the slate implementation is finished.. 
13:22:14 <water> and primitive MOs are ones written and compiled in C++ :)
13:22:46 <Ghyll> what I'm interested in, is writing my own mo for array access, and expect it to be as fast as can be.
13:22:55 <Ghyll> ie, as fast as native C code. 
13:23:01 <water> if it's finished, then the whole implementation is written in slate and compiles to native where needed (or to multiple ILs)
13:23:45 <water> the idea is to have a fluid range of performance where you pick the point you want in a reflective interface
13:24:16 <Ghyll> what do I have to trade to get performance ?
13:25:13 <water> compile time
13:25:30 <water> less interactivity over the span of the compile
13:25:49 <water> (which will likely occur in a separate thread)
13:28:08 <Ghyll> can I extend the compiler inside the language ? 
13:28:21 <water> once it's self-hosted, of course
13:28:55 <Ghyll> I really hope all this can be pulled off in a timely manner. 
13:29:04 <water> we'll see
13:29:22 <water> maybe Fare's new company will do some of the heavy work
13:29:24 <Ghyll> I've mentioned it before and I mention it again, I'm willing to spend some time helping out with the implementation once the docs are more complete.
13:29:28 <water> maybe not :)
13:30:43 <water> ok
13:31:05 <Ghyll> I'm in the fortunate position of working in a company that understands the open source philosophy. they've given us a go ahead to spend a few man months on adding functionality to postgresql because we need it. 
13:31:16 <water> in that case, once pidgin is ready to use, then docs will be created
13:31:32 <water> cool
13:31:34 <Ghyll> if I could convince the other developers we need slate, I guess I could get permission to work on it full time a day or two a week.
13:31:48 <water> "need slate"?
13:31:58 <Ghyll> although I'd have to come up with a project where other languages don't fit the bill.
13:31:59 <water> that's a bit of a stretch
13:32:04 <Ghyll> yes, I now :)
13:32:08 <Ghyll> now/know
13:33:30 <Ghyll> alternatively, I could make it my master thesis project to implement slate, but I don't think that would be proper. but at least I'd be available to do work on slate full time for about a year.. 
13:33:45 <water> hm
13:33:48 <smklsmkl> did somebody read robert glück's papers?
13:33:58 <water> i also am going back to school
13:34:08 <water> smkl: i forget, which are those?
13:34:08 <Ghyll> smklsmkl: no. url ? subject ?
13:34:47 <Ghyll> anyway. it's getting late over here, supper is long overdue. 
13:34:52 <water> ok
13:35:05 <Ghyll> we'll talk later. 
13:35:08 [QUIT] Ghyll quit: foo
13:35:09 <water> sure
13:35:50 <smklsmkl>  http://www.diku.dk/research-groups/topps/bibliography/1994.html#D-185
13:36:08 <water> oh those
13:36:13 <water> i haven't read them all yet
13:38:40 <smklsmkl> do they have anything interesting?
13:39:06 <water> nothing earth-shaking, but yes
13:57:26 <water> what's with abi's amnesia?
13:57:37 <ult> She's broken
13:57:41 <ult> hcf didn't have time to fix her
13:57:51 <water> hcf was here?
13:57:57 <ult> yep
13:58:07 <ult> last night
13:58:10 <water> where has he been?
13:58:22 <ult> for about 15 seconds, saying abi was broken and would stay so for the time being.
14:04:37 [QUIT] ult quit: Leaving
14:24:13 [QUIT] gREMLiNs quit: BitchX: coming soon to a theatre near you!
15:20:01 [QUIT] water quit: The Tao went that-a-way!
16:31:39 water joined #tunes
17:11:13 lar1 joined #tunes
17:12:26 [QUIT] smklsmkl quit: back tomorrow
18:28:32 ult joined #tunes
19:00:37 eihrul joined #tunes
22:14:55 frogman joined #tunes
23:46:19 water joined #tunes
23:46:42 <water> re all
23:46:52 <water> oo... an amphibian! :)
23:47:02 <water> hi frogman
01:30:22 moebious joined #tunes