00:00:00 --- log: started retro/13.07.26
00:56:12 --- quit: Kumool (Quit: Divided by 0)
04:20:24 --- quit: Sgeo (Read error: Connection reset by peer)
08:46:21 --- join: ncv (~quassel@79.114.115.190) joined #retro
08:46:21 --- quit: ncv (Changing host)
08:46:21 --- join: ncv (~quassel@unaffiliated/neceve) joined #retro
09:13:45 --- join: Mat2 (~claude@91-65-144-133-dynip.superkabel.de) joined #retro
09:13:50 <Mat2> hi @ all
09:14:37 --- join: Kumul (~nmz@adsl-64-237-237-61.prtc.net) joined #retro
09:15:57 <tangentstorm> heya
09:16:14 <Mat2> hi tangentsorm !
09:16:56 <Mat2> ^tangantstorm, sorry
09:20:10 <tangentstorm> i get the idea :)
09:30:39 <tangentstorm> working on a tool to convert diagrams from that VUE application to records in a database and then svg
09:31:11 <tangentstorm> vue == http://vue.tufts.edu ... can't remember who i already talked about this with
09:31:59 <tangentstorm> right now i'm just using sqlite but my goal is to get the data into ram inside retro/ngaro
09:34:29 <Mat2> ah ok
09:35:58 <Mat2> have you finsihed your pascal-to-retro compiler ? I'm asking because native-code generation is on the way and I plan to extend it a bit for compiling Nimrod
09:38:54 <tangentstorm> nope
09:39:18 <tangentstorm> i have most of the templates worked out
09:40:44 <tangentstorm> i found i have so many pieces to my project that i needed to refocus and start working from the top down, breadth first :)
09:41:04 <tangentstorm> otherwise i just get lost in the details and jump around from one part to another
09:41:17 <Mat2> hmm, ok I will help you out after finishing the compiler
09:41:27 <tangentstorm> :)
09:42:02 <tangentstorm> compiler for saiwa? retro? nimrod?
09:43:43 <Mat2> I extend my natice-code compiler for Saiwa with a frontend which compiles Parable slices via translating into AVm as IL layer
09:44:28 <tangentstorm> :D
09:44:47 <tangentstorm> slices meaning dictionary entries?
09:45:54 <Mat2> no, the memory is addressed in pieces of a fixed size and each defination compiles into one of them
09:46:27 <Mat2> because Parable is quite similar to retro in most ways, porting your compilier is trivial
09:47:54 <Mat2> my compiler rpelaces the Parable VM, your compiler code used as base for compiling Nimrod and the result is a Nimrod native-code compiler which can run on the duinomite :)
09:47:58 <Mat2> that's my idea
09:48:08 <Mat2> ^replaces
09:55:54 <Mat2> I think so all fitting together
10:01:51 * Mat2 code-compilation works now
10:03:12 <crc> rx-core.org differs from retroforth.org when I'm working on major changes to retro. (the name was an abbreviation for retro experiemental core)
10:03:38 <Mat2> hi crc !
10:05:50 <crc> hi mat
10:06:41 <Mat2> crc: By the way, I think Parable need an abstraction for organizing definations
10:06:56 <Mat2> at code level
10:09:13 <crc> can you explain>
10:12:51 <Mat2> Forth features vocabularies for namespace seperation
10:13:29 <Mat2> and this also simplifies object-oriented extensions
10:16:24 <Mat2> there are many possible demands I can think of where organizing definations by seperations helps programming
10:16:56 <Mat2> how about seperate compilation in parable ?
10:17:07 <Mat2> ^separate
10:26:32 <Mat2> hmm, make it sense to support instruction combinations which duplicate TOS and then left or right-shift ?
10:29:31 <tangentstorm> depends what you're doing i guess
10:29:51 <tangentstorm> if you wanted to pack multiple values into the cell maybe
10:30:06 <tangentstorm> like extract a particular byte...
10:31:25 <crc> I've held off on vocabularies so far as I'm still considering how to handle them. I'm not quite happy with the way they turned out in retro, so am still considering what to do.
10:34:12 <Mat2> tangentstorm: ah yes! For extracting scalar values after computing SIMD operations that's of use
10:34:20 <Mat2> thnaks
10:35:05 <Mat2> crc: what's your problem with the retro implementation ?
10:35:10 <Mat2> ^ thanks
10:39:01 <crc> Mat: there have been a number of odd bugs in it in the past (causing dictionary corruption and unusable images). mostly fixed now, but I tend to use the ^ prefix to avoid multiple open vocabularies for this reason.
10:39:50 <crc> it's also difficult to modify a vocabulary after it is sealed with ;chain
10:40:18 <tangentstorm> hrm... why, crc?
10:43:17 <Mat2> crc: I think the main reason for this difficulties are simply side-effects of the linked-list approach and should be avoidable using an index referenced database of fixed hash-tables
10:43:19 <tangentstorm> if you have the last entry handy, couldn't you have a little word that does:   @last chain-last !last
10:43:25 <crc> retro was only intended to have a single vocabulary. so each vocabulary is technically a completely separate dictionary chain
10:44:01 <crc> there probably should be a function to move the most recent header into a specified chain, but I haven't actually implemented or tested this yet
10:44:34 <tangentstorm> oh that would be cool :)
10:44:48 <tangentstorm> almost like a refactoring tool
10:45:23 <crc> it's not quite as easy to pull off in parable since the dictionary is not exposed via the instruction set or memory slices
10:46:34 <Mat2> <- that's why I think the isntruction-set should be extended providing these functionality
10:46:40 <Mat2> ^instruction
10:49:32 <tangentstorm> hate to do a Mat2 impression but i need to get some sleep. see you guys :)
10:49:38 --- quit: tangentstorm (Quit: WeeChat 0.3.2)
10:49:58 <Mat2> ciao
10:51:00 <Mat2> hmm, probably a ordered tree would be easier to implement
10:52:47 <Mat2> crc: I think the solution would be support of some kind of database for this
10:53:16 <Mat2> a really simple one of course
10:55:13 <Mat2> the web-based versions can then simply map the instructions to the JSON protocol probably
11:09:28 <Mat2> crc: so my suggest is: Use a relational-database approach (remember, I had some ideas before in this direction) for Parable
12:12:33 <crc> my eventual goal is to build a vm that uses sqlite to store everything
12:25:38 <Mat2> that's nice but would not help building Parable fro small systems - but eventually a subset
12:25:44 <Mat2> ^for
12:56:59 <crc> true, but my current vm implementations won't run on a small system anyway
12:58:21 <Mat2> then let's if I can recode the routines you will use
12:58:42 <Mat2> otherwise I it out
12:58:49 <Mat2> ^I left
14:36:51 <Mat2> ciao
14:37:04 --- quit: Mat2 (Quit: Verlassend)
16:26:23 --- join: Sgeo (~quassel@ool-ad034ea6.dyn.optonline.net) joined #retro