10:58 < heinrich5991_> matricks: in case of an interpreted script, would it be sufficient if it advertised its own max memory usage (stack only) and maximum instruction count?
10:59 <@matricks> you could do stuff like that
11:00 <@matricks> the thing I like about small is that it knows the memory usage from start apparently
11:00 <@matricks> I havn't looked into how it does it however
11:01 < heinrich5991_> well, it says it knows it when it doesn't recurse, and this way you can (simply) calculate its maximum stack usage by considering all calls to be taken
11:01 <@matricks> yeah
11:02 < heinrich5991_> the instr count problem still stands though
11:02 < heinrich5991_> but it seems (from what I can tell) that pawn would allow for segfaults
11:02 <@matricks> I solved that issue in my language by only allowing to skip instructions, not jumping back :)
11:03 <@matricks> heinrich5991_: they perform bounds checking and you will get an error back I think
11:03 < heinrich5991_> yea, but that even disallows the simplest loops, such as for player-player collision
11:03 <@matricks> heinrich5991_: this was ment for rendering of stuff
11:04 <@matricks> heinrich5991_: and you can handle the cases where you have to loop by making the real gamecode call the functions for every p2p collision resolve etc
11:04 < heinrich5991_> the manual said something about being careful with non-initialized memory as strings are zero-terminated and when printed could cause a segfault
11:05 <@matricks> oh-shi...
11:05 <@matricks> dunno, I havn't used it for anything or looked into it that much
11:06 < heinrich5991_> it seems that pawn isn't very much documented, got that from a tutorial
11:06 < heinrich5991_> https://wiki.alliedmods.net/Introduction_to_sourcepawn#Caveats
11:07 < heinrich5991_> I didn't find a documentation for the bytecode yet
12:04 <@matricks> heinrich5991_: that caveat seems quite bad :/
20:09 < heinrich5991> matricks: you said that you created some language with forward-only jumps
20:11 < heinrich5991> do you still have the bytecode reference or whatever you compiled to?
20:26 <@matricks> heinrich5991: yes
20:27 < heinrich5991> I'd like to read it, if that was possible
20:29 <@matricks> I can dig it out a bit later, a bit busy right now
20:30 < heinrich5991> ok, thank you :)
20:34 < Dunedune> matricks: whoa that's interesting, what are its purposes?
20:34 <@minus> still working matricks?
20:35 <@minus> start work at half past 6, go home before everyone else :D
21:34 <@matricks> Dunedune: experiment with sending code to the client for rendering
21:34 <@matricks> minus: no, cleaning appartment, running errends etc
21:35 <@matricks> heinrich5991: going through backups now
21:35 <@matricks> heinrich5991: found it
21:36 < heinrich5991> that was quick
21:36 <@matricks> I'm pro like that
21:36 < heinrich5991> ordered backups :)
21:36 <@matricks> heinrich5991: not really
21:36 <@minus> well-organized backups, nice
21:36 <@matricks> just looked through the latest backup
21:36 < PsychoGod> mmm, something interestid is going on :3
21:37 <@matricks> http://matricks.se/vm/vm.tar.gz
21:37 <@matricks> http://matricks.se/vm contains all the files as well
21:37 <@matricks> compiler.py == the compiler
21:38 <@matricks> test.cpp + opcodes.h is the virtual machine + test
21:38 <@minus> such macro
21:38 <@minus> looks /very/ minimal
21:38 <@matricks> it is
21:39 <@matricks> but it should almost do everything that is needed for the client to render game objects
21:39 <@matricks> and you can easily extend it
21:40 <@matricks> compiler == 550 lines of python :)
21:41 <@matricks> erh.. you need the CBurkVMBuilder in test.cpp as well yeah.
21:43 <@matricks> okey.. the test.cpp does some other stuff as well
21:43 <@matricks> it uses macros and some fancy c++ to run functions as produce vm bytecode for em :D
21:43 <@matricks> haha, I don't remember doing that
21:45 <@minus> "i must've been drunk"
21:45 <@matricks> that was kinda cool :)
21:46 <@matricks> but yeah, the actual virtual machine is like.. 100 lines of code
21:46 <@matricks> 150..
21:47 <@matricks> and there is a way to crash it ;)
21:47 <@matricks> perhaps...
21:47 <@matricks> 5 points to the one who can figure it out
21:48 <@minus> call 0
21:48 <@matricks> call doesn't do anything
21:48 <@minus> aw
21:49 <@matricks> okey, you can't crash it :)
21:49 <@matricks> floating point exceptions are not enabled by default
21:55 <@matricks> it's kinda fun how little code you need todo a small vm with compiler these days
21:56 < EastByte> thanks to python
21:56 <@matricks> among other things
21:59 <@matricks> and people need to stop competeing in paragliding if Maurer is involved :D
21:59 <@matricks> Maurer == A god in the paragliding world
22:00 <@matricks> he have won the hardest, most challening competition 3 times in a row now
22:01 < EastByte> wow that's great :)
22:01 < EastByte> now continue working on cool projects :D
22:02 <@matricks> I'm gonna finish up my cleaning of my appartment
22:02 < EastByte> lol how long does this take?
22:02 <@matricks> 1 more hour perhaps
22:02 < EastByte> ...
22:02 <@matricks> been cleaning the balcony etc
22:02 <@matricks> cleaning out some closests
22:03 <@matricks> too many gay people in them
22:03 <@matricks> and I spent loads of money today \o/
22:03 < PsychoGod> D: it's not good with gay people
22:03 <@matricks> ordered my tandem gear
23:20 <@matricks> aaaand... almost done
23:20 <@matricks> :)
23:21 < heinrich5991> matricks: how many registers does your machine have?
23:21 <@matricks> heinrich5991: the vm?
23:21 < heinrich5991> yes
23:21 <@matricks> none
23:22 <@matricks> registers == memory more or less
23:22 < heinrich5991> move r4, r6
23:22 < heinrich5991> aha
23:22 <@matricks> you don't have registers and memory.. you just have registers.. or memory :) it's called both :)
23:22 < heinrich5991> so you have only-addressable-by-constant memory?
23:23 <@matricks> at the time I wrote that, yes
23:23 <@matricks> you could easily add an indirect move
23:23 <@matricks> or "load"
23:24 <@matricks> MACRO_OPCODE(LOAD,REG,REG,NON,mem[p0] = mem[mem[p1]], "load r%u, r%u") // load rDEST, r[rSOURCE]
23:24 < heinrich5991> in theory, could one use LLVM to compile to one's bytecode?
23:24 <@matricks> there.. done
23:24 <@matricks> heinrich5991: "yes"
23:24 <@matricks> heinrich5991: and no
23:24 < heinrich5991> matricks: what you wrote is still unsafe
23:24 <@matricks> as soon as you get a jump instruction, you are bone
23:24 <@matricks> boned
23:25 <@matricks> heinrich5991: nope
23:25 <@matricks> heinrich5991: it's safe
23:25 < heinrich5991> is mem a class?
23:25 <@matricks> oh wait...
23:25 <@matricks> 2 sec
23:26 <@matricks> MACRO_OPCODE(LOAD,REG,REG,NON,mem[p0] = mem[mem[((size_t)p1)%CBurkVM::MEMORY_SIZE]], "load r%u, r%u") // load rDEST, r[rSOURCE]
23:26 <@matricks> there
23:26 <@matricks> or.. is size_t signed? :)
23:26 < heinrich5991> size_t is unsigned
23:26 <@matricks> then there :)
23:27 < heinrich5991> why not abort the VM on out of memory access?
23:27 < heinrich5991> have you prevented div by 0 btw?
23:27 <@matricks> heinrich5991: div by 0 -> inf
23:28 <@matricks> heinrich5991: you could halt execution if you want
23:28 <@matricks> heinrich5991: I just made an example
23:28 <@matricks> wait.. I did that opcode wrong :D
23:29 <@matricks> MACRO_OPCODE(LOAD,REG,REG,NON,mem[p0] = mem[((size_t)mem[p1])%CBurkVM::MEMORY_SIZE]], "load r%u, r%u") // load rDEST, r[rSOURCE]
23:29 <@matricks> there
23:29 <@matricks> bleh
23:30 <@matricks> it will have some issues however :D
23:30 <@matricks> as you are casting floats to ints
23:30 < heinrich5991> well, you don't have enough registers to cause strange things I guess
23:31 <@matricks> the thing isn't really tested or so, just made as a small test
23:32 <@matricks> hmm, I should finishing watching that movie I started watching yesterday :D
23:34 < Elzeph> matricks, what's the title?
23:34 <@matricks> Nymphomanic
23:35 <@matricks> got Vol II left to see
23:35 <@matricks> stupid long movie :) 4 hours.. and that's the short version