Thu Sep 17 21:38:00 2009
[imud_code]
Ideysus@ShadowMUDii: But people probably won't like me just stripping out the logging in error_handler
Thu Sep 17 21:39:22 2009
[imud_code]
Kalinash@Fire and Ice: why would you not want the logs? if a catch() fails it's a BadThing (tm)
Thu Sep 17 21:40:07 2009
[imud_code]
Ideysus@ShadowMUDii: Okay, as an example read_file in grep would die on binary files
Thu Sep 17 21:41:55 2009
[imud_code]
Kalinash@Fire and Ice: why do you have binary files? you can't create binar\y files from within the mud
Thu Sep 17 21:43:37 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. Does catch() change at all in fluffos?
Thu Sep 17 21:43:56 2009
[imud_code]
Kalinash@Fire and Ice: no... but the behavior you're talking about is controlled by the mudlib
Thu Sep 17 21:45:01 2009
[imud_code]
Kalinash@Fire and Ice: the error_handler apply in the master object
Fri Sep 18 04:05:34 2009
[imud_code]
Ideysus@ShadowMUDii: So. How feasible is it to use present() in a custom parser?
Fri Sep 18 04:06:33 2009
[imud_code]
Ideysus@ShadowMUDii: I believe present() tries to be somewhat clever yes? Like, you can do things like 'orc 2'? Is that the limit of it?
Fri Sep 18 04:08:05 2009
[imud_code]
Ideysus@ShadowMUDii: I'm just a tiny bit fearful of the impact on the lib of a parser based around 'filter(objs, (:id=="whatever":))'
Sat Sep 26 19:51:58 2009
[imud_code]
Ideysus@ShadowMUDii: I'd sell my soul for a {: :} construct which worked like (: :) except allowed statements instead of just an expression.
Sat Sep 26 19:55:02 2009
[imud_code]
Cratylus@Dead Souls Dev: incidentally, "two beers" in japanese is "ni biru"
Sat Sep 26 20:21:27 2009
[imud_code]
Silenus@Dead Souls Dev: cant you just use the function(int x,int y) {} syntax?
Sat Sep 26 20:23:08 2009
[imud_code]
Aidil@Way of the Force: no syntax is a replacement for two beers! :)
Sat Sep 26 21:02:27 2009
[imud_code]
Qwer@Archipelago: ideysus: you might not have to sell your soul just yet
Sat Sep 26 21:02:33 2009
[imud_code]
Qwer@Archipelago: int inc(int x) { function f = function(int y) { return y + 1; }; return evaluate(f, x); }
Sat Sep 26 21:03:10 2009
[imud_code]
Qwer@Archipelago: err, which was what silenus was getting at :)
Sat Sep 26 21:03:37 2009
[imud_code]
Cratylus@Dead Souls Dev: i am still quite certain that beer would improve whatever situation is being discussed
Sat Sep 26 21:03:59 2009
[imud_code]
Qwer@Archipelago: for once in my life, i'm fresh out of beer :(
Sat Sep 26 22:45:11 2009
[imud_code]
Ideysus@ShadowMUDii: The function(){} syntax isn't a closure though.
Sat Oct 3 21:01:14 2009
[imud_code]
Silenus@Dead Souls Dev: is it possible to find out the cost of L1 versus L2 versus L3 cache latencies?
Sat Oct 3 21:02:31 2009
[imud_code]
Silenus@Dead Souls Dev: is it something you have to look up for the specific architecture?
Sat Oct 3 21:04:38 2009
[imud_code]
Sys@BlackHole: tho there migth be a way to test for it on 'this' system
Sat Oct 3 21:05:16 2009
[imud_code]
Silenus@Dead Souls Dev: *nod* just curious how you would go about it.
Sat Oct 3 21:05:35 2009
[imud_code]
Silenus@Dead Souls Dev: just trying curious what a ballpark figure might be too
Sat Oct 3 21:05:58 2009
[imud_code]
Sys@BlackHole: now those questions are much trikier than your earlier ones. soz but fiik
Thu Oct 15 00:31:32 2009
[imud_code]
Ideysus@ShadowMUDii: Does fluffos have the annoying 'Function pointer returning string constant is NOT a function call' warning?
Thu Oct 15 00:34:07 2009
[imud_code]
Cozminsky@EotL: ldmud returns #'usr/cozminsky/cc->__inline_usr_cozminsky_cc_c_3_0000()
Thu Oct 22 19:28:43 2009
[imud_code]
Kejope@Cookie Dough: Subject: Compiler/language creation. What is your opinion of a single source file containing code at the top, and non-code text at the bottom, separated by a token at the end of the code part, i.e., ENDCODE, such that the compiler would treat the ENDCODE as EOF, completely ignoring anything afterwards? So, my BNF might include something like, <file> ::= <code> NEWLINE 'ENDCODE' NEWLINE CHAR*, where the <code> would be everthing that <file> usually would be. I understand that I could just enclose the area within a comment, but I would like to avoid having the compiler waste any time processing this area.
Thu Oct 22 19:29:34 2009
[imud_code]
Kejope@Cookie Dough: Silenus, are you asking me if BASH can read config files? I think I am misunderstanding you.
Thu Oct 22 19:29:49 2009
[imud_code]
Kejope@Cookie Dough: Yes. Glutton for challenging punishment. :)
Thu Oct 22 19:32:06 2009
[imud_code]
Silenus@Dead Souls Dev: to use lex/yacc or antlr (if speed isnt a premium)
Thu Oct 22 19:32:17 2009
[imud_code]
Kejope@Cookie Dough: Well, the biggest pain was being able to import a file without bash clobbering the raw text. In this project, one of my personal goals is to use as few "outside" utilities as possible. Yes, I have a small library of bash routines now that allow me to raw-read a file, then process it however from there on as a string variable.
Thu Oct 22 19:32:51 2009
[imud_code]
Silenus@Dead Souls Dev: well i dont know enough about bash to say unfortunately :S
Thu Oct 22 19:32:54 2009
[imud_code]
Andrew@Nanvaent: What was wrong with cat <<EOF ; <raw text> ; EOF ??
Thu Oct 22 19:34:09 2009
[imud_code]
Kejope@Cookie Dough: cat will convert newlines to spaces. I also did not want bash converting any t chars for me. ..and so forth.
Thu Oct 22 19:37:55 2009
[imud_code]
Kejope@Cookie Dough: I do like the idea of lex/yacc, but I started this project to see what I could do with bash. I am not complaining about its limitations (at least not in this conversation :> ). I acknowledge it is not the best way for many reasons. One of which is its speed. The slowest part at the moment is reading in the file. If I could just stop reading in upon hitting a specific token, I could gain some significant speed. I am sort of just wondering about the merit of using such a token, or whether there are some problems with doing such a thing, that I just have not thought of yet.
Thu Oct 22 19:38:55 2009
[imud_code]
Kejope@Cookie Dough: Yes, I do frequently find it useful to put notes and sample code and data at the bottom of a source file while I am developing. :)
Thu Oct 22 19:40:43 2009
[imud_code]
Andrew@Nanvaent: And if you just want the stuff from the file do something like awk -e 'show == 1 { print; } /^exit 0$/ { show = 1; }'
Thu Oct 22 19:42:12 2009
[imud_code]
Kejope@Cookie Dough: Using exit 0 feels more like misusing the language somehow. Maybe I just need to change my perspective on that part. :> I know I can use awk, or any number of other utilities. So far, I am making lots of progress without having a single awk or sed in my code, or most any other common unix program.
Thu Oct 22 19:43:55 2009
[imud_code]
Kejope@Cookie Dough: I know I have limited myself with my requirements, but it is a useful exercise for me. "I'm going to duel him with my left hand." "You know what a hurry we're in!" "It's the only way I can be satisfied, otherwise, it's over too quickly." :)
Thu Oct 22 19:50:02 2009
[imud_code]
Kejope@Cookie Dough: I almost didn't post my question, because I just knew the focus would be to try to help me see the error of my ways with bash. hehe I appreciate all of your thoughts. I know well the feeling of, "Why is the idiot doing it THAT way?" I just wondered if the ENDCODE idea had pros or cons, in principle, that I wasn't considering.
Thu Oct 22 19:56:48 2009
[imud_code]
Sys@BlackHole: surely the big issue is if you want ENCODE to appear somewhere else in the program
Thu Oct 22 19:57:13 2009
[imud_code]
Sys@BlackHole: in other wordss you need to choose what ENDCODE is carefully
Thu Oct 22 20:01:20 2009
[imud_code]
Sys@BlackHole: I spose if you prefix it with the comment char then you just have to remember to not start any comment with ENDCODE
Thu Oct 22 20:01:42 2009
[imud_code]
Kejope@Cookie Dough: The compiler will never even get to the end of the code part to see the token, if there is any syntax error in the code. Following my BNF definition, the compiler would never even try to parse the characters 'ENDCODE' as anything special unless it has already successfully parsed the code, then also parsed a newline, then 'ENDCODE' on a line by itself, then another newline. In this case, I should be able to use almost any word I like.
Thu Oct 22 20:03:16 2009
[imud_code]
Kejope@Cookie Dough: Even a comment block, by definition, would only be recognized as a comment block, inside the code block area. http://en.wikipedia.org/wiki/Backus%E2%80%93Naur_Form Or am I misunderstanding about how BNF works? I have a lot of my compiler working now, and it seems to follow the BNF patterns nicely.
Thu Oct 22 20:04:51 2009
[imud_code]
Silenus@Dead Souls Dev: are you just trying to look for end token?
Thu Oct 22 20:05:27 2009
[imud_code]
Silenus@Dead Souls Dev: or actually trying to parse something (and beyond that compile it)
Thu Oct 22 20:07:24 2009
[imud_code]
Kalinash@Fire and Ice: all this token' is why you never get your work done
Thu Oct 22 20:07:25 2009
[imud_code]
Silenus@Dead Souls Dev: and i am actually quite curious how you managed to parse something in BNF form in bash without reinventing the wheel
Thu Oct 22 20:08:26 2009
[imud_code]
Silenus@Dead Souls Dev: i.e. having to build some sort of LL(k) LR(k) or packrat style parser
Thu Oct 22 20:13:11 2009
[imud_code]
Ideysus@ShadowMUDii: Wait... you are coding a programming language parser in Bash?
Thu Oct 22 20:14:19 2009
[imud_code]
Kejope@Cookie Dough: I suspect that the pre-defined, self-imposed limitations of my project are muddying the focus of my question. I apologize. Let me try to start over. Most importantly, I already have the technical capability in my compiler to use an ENDCODE token to separate the top and bottom portions of my source file. My question is more along the lines of, do you think this 'code and non-code' combination in one file is a good idea or bad idea, or perhaps just a silly, redundant comment type? I can imagine how it could be taken to extremes, to have binary data in the same file as the source code. I don't think that is such a useful concept at the moment (it really defeats the point of source code as opposed to a compiled binary, for one). To answer your most recent questions, I am trying, and succeeding in, parsing source code for compilation. I am sort of reinventing the wheel. The process has helped my programming skills nicely. Yes, this is an LL(1) compiler. Yes, I am insane, thank you! :)
Thu Oct 22 20:14:27 2009
[imud_code]
Ninja@Dead Souls Dev: but, if he succeeds, I'm a gonna use that programming language parser to code up a bash-emulator.
Thu Oct 22 20:15:28 2009
[imud_code]
Ninja@Dead Souls Dev: history lesson...in the OLDE days, code AND data were both contained in the same source file.
Thu Oct 22 20:17:33 2009
[imud_code]
Silenus@Dead Souls Dev: I guess it might be an interesting exercise :D
Thu Oct 22 20:17:48 2009
[imud_code]
Ideysus@ShadowMUDii: Doesn't every source file include code and data in one file? The question is just how bound up the data is in the code. In some of the perverse cases I suppose the code is bound up in data...
Thu Oct 22 20:17:52 2009
[imud_code]
Sys@BlackHole: however I think seperating code and data is saner, se Template::Toolkit for example
Thu Oct 22 20:18:34 2009
[imud_code]
Kalinash@Fire and Ice: 10 RESTORE:FOR X=1 TO 4:READ S$:PRINT S$;:NEXT X
Thu Oct 22 20:18:47 2009
[imud_code]
Sys@BlackHole: that way you can hand the data portion to someone who deosn't do code, and they aren't scared
Thu Oct 22 20:21:22 2009
[imud_code]
Sys@BlackHole bows down before Silenus, for if he has really hacked Kali good he is GOD!
Thu Oct 22 20:21:44 2009
[imud_code]
Silenus@Dead Souls Dev: hmm i thought i hacked the checksums too :D
Thu Oct 22 20:21:56 2009
[imud_code]
Kejope@Cookie Dough: Ninja, yes, I should have thought of that viewpoint. It's been too long since I actively worked in Assembly or BASIC, I guess. So, that really does answer my question in a sense that I am familiar with. Although I actually have no intention to use the area for data, just for docs and notes and such. I will now look into Template::Toolkit, Sys. Thanks.
Thu Oct 22 20:23:45 2009
[imud_code]
Sys@BlackHole: if it's just for docs and notes and that how does it differ from comments? and surely commenting inline is a better bet as you can doc the code where it needs docing, there's also nothing to stop you having big comments at the beginning or end, which is common in bash files in my experience
Thu Oct 22 20:27:00 2009
[imud_code]
Sys@BlackHole: I'm also not sure that Template::Toolkit will help much, but it is a method of writing cgis where the html and code (perl) are in seperate files, so the designer doesn't get scared by the perl magic. It has other uses too.
Thu Oct 22 20:29:58 2009
[imud_code]
Kejope@Cookie Dough: Sys, now we are getting back to the meat of why I considered an ENDCODE token! It is agreed that bash is not the most efficient language for a compiler. After much research, I have learned that bash can not easily read in a file without twisting the text stream in its own nefarious ways. I have used a combination of while read -r line, and manually inserting tokens into the stream for newline, and for the token's starting character (a tilde for now). My resulting bash variable is a faithful copy of the disk file, with the exception that it has tokens. Other functions allow me to work with that variable similarly to a C++ stream, automatically taking care of the token conversion. When I get to a big file, this main file reading function appears frozen for unpleasantly long periods of time. My thought was to stop reading in the source file at ENDCODE, thereby saving significant amounts of time while recursing through many source files.
Thu Oct 22 20:30:19 2009
[imud_code]
Kalinash@Fire and Ice: ok, you need to limit your chat to short, witty comments
Thu Oct 22 20:33:56 2009
[imud_code]
Detah@Dead Souls Dev: Im afraid I cannot comply with that request either, K.
Thu Oct 22 20:34:08 2009
[imud_code]
Kejope@Cookie Dough: Short and witty, eh? Ah. It reminds me that perhaps not everyone is using a 160 char wide terminal. :>
Thu Oct 22 20:45:07 2009
[imud_code]
Sys@BlackHole: heresy !!! everyone knows terminals are 80chars wide, but if speed is your goal why write in bash?
Thu Oct 22 20:50:36 2009
[imud_code]
Kejope@Cookie Dough: Sys, I definitely doc inline! These notes are sometimes more ... general notes. Also, the actual code in the source files look ... ugly, for now. I have been trying to keep those areas with a cleaner look, while I am debugging my compiler. Ultimately, the notes I am referring to, probably should really go somewhere else, outside of the source file -- but it is so convenient to be in the same file, for now.
Thu Oct 22 20:51:32 2009
[imud_code]
Sys@BlackHole: sounds like you need a bugtracking system with a wiki, like Trac say, but each to thier own
Thu Oct 22 20:52:40 2009
[imud_code]
Kejope@Cookie Dough: Speed is not one of the main design goals. I really have only one issue with speed, that is, reading in a file. I think that an ENDCODE token helps resolve that. I am just wondering if it is a meritus idea for future use.
Thu Oct 22 20:55:24 2009
[imud_code]
Kejope@Cookie Dough: I have considered bugtracking software, and do like the idea, but have not used them before. I am trying (somewhat successfully) to contain the scope of my immediate project. It partly started as a project to practice with git. I think I don't want to try to make it a full, professional development environment project at this time. :)
Thu Oct 22 20:57:12 2009
[imud_code]
Kejope@Cookie Dough: Based on this conversation, I have decided to just use the ENDCODE token as a tiny, short-term hack. The general resistance to the idea seems to indicate that the idea is not so popular. Cool! :)
Thu Oct 22 20:57:55 2009
[imud_code]
Ninja@Dead Souls Dev: problem solved...let's work on world hunger next./
Thu Oct 22 20:59:04 2009
[imud_code]
Sys@BlackHole: I think there's probably more resistance to the idea of writing a compiler in bash
Thu Oct 22 21:08:20 2009
[imud_code]
Silenus@Dead Souls Dev: do you guys know interprocedural data flow tricks? I have this idea and i think it's workable for LPC
Thu Oct 22 21:11:42 2009
[imud_code]
Silenus@Dead Souls Dev: i probably couldnt but maybe you could :D
Thu Oct 22 21:14:05 2009
[imud_code]
Kejope@Cookie Dough: I am not so great. I just like to learn. I am not expert on data flow, but if you are willing to discuss the idea, I am almost certain that we would be happy to beat you up fo---- I mean, happy to be sounding boards. :)
Fri Oct 23 04:03:42 2009
[imud_code]
Ideysus@ShadowMUDii: Is there such a concept as a multidimensional grammar?
Fri Oct 23 04:05:21 2009
[imud_code]
Ideysus@ShadowMUDii: So, like maybe you get S=>AB/CD (that is, a square with AB above CD)
Sun Oct 25 22:33:22 2009
[imud_code]
Silenus@Dead Souls Dev: hmm I have been wrestling with my struct/classes compat designs again and havent found a solution which is particularly satisfactory
Sun Oct 25 22:33:45 2009
[imud_code]
Silenus@Dead Souls Dev: would it make any sense to treat them as kind of type bags?
Sun Oct 25 22:34:45 2009
[imud_code]
Silenus@Dead Souls Dev: I thought about making any struct declaration global to the system but I am a bit concerned that infering
Sun Oct 25 22:35:14 2009
[imud_code]
Silenus@Dead Souls Dev: the order in which things load may affect things in wierd ways
Sun Oct 25 22:36:16 2009
[imud_code]
Silenus@Dead Souls Dev: i can map them into real struct like structures since LLVM supports that
Sun Oct 25 22:37:14 2009
[imud_code]
Silenus@Dead Souls Dev: but the question becomes if the ordering of fields in the declaration should or should not be ignored
Sun Oct 25 22:38:35 2009
[imud_code]
Silenus@Dead Souls Dev: is the correct view to sort of look at a struct declaration as a sort of typedef for a tuple?
Sun Oct 25 22:39:08 2009
[imud_code]
Silenus@Dead Souls Dev: i think this is how it works in fluffos atm
Sun Oct 25 22:39:46 2009
[imud_code]
Silenus@Dead Souls Dev: which are structs for all intents and purposes
Sun Oct 25 22:44:22 2009
[imud_code]
Silenus@Dead Souls Dev: i am just wondering if struct A { int blah , string blah2 } should be the same as struct B { string blah 2, int blah }
Sun Oct 25 22:45:45 2009
[imud_code]
Silenus@Dead Souls Dev: or if they should be distinct type wise if you throw away all the labels
Sun Oct 25 22:46:11 2009
[imud_code]
Cratylus@Dead Souls Dev: only slightly awokened. still dealin with everyone bein sick
Sun Oct 25 22:46:31 2009
[imud_code]
Ideysus@ShadowMUDii: I didn't think they had a concept of order in lpc.
Sun Oct 25 22:47:02 2009
[imud_code]
Silenus@Dead Souls Dev: internally they are mixed arrays of fixed size
Sun Oct 25 22:47:19 2009
[imud_code]
Ideysus@ShadowMUDii: Though come to think of it I've never used them. I just use an array and a list of #defined offsets
Sun Oct 25 22:47:40 2009
[imud_code]
Silenus@Dead Souls Dev: so once the compiler is done with it the type information is discarded
Sun Oct 25 22:48:28 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. Its all done at compile time? So what happens if you try to use structs in distinct objects?
Sun Oct 25 22:48:44 2009
[imud_code]
Silenus@Dead Souls Dev: i am trying to tighten it up a bit and retain the type informatio internally
Sun Oct 25 22:50:40 2009
[imud_code]
Silenus@Dead Souls Dev: my suspicion is that you could do some wierd stuff with it which might be slightly counterintuitive but i havent tried it.
Sun Oct 25 22:52:26 2009
[imud_code]
Silenus@Dead Souls Dev: how best to make structs a real type in the system
Sun Oct 25 23:02:11 2009
[imud_code]
Wodan@Discworld: could add a field to the array used for classes to point to some unique class description
Sun Oct 25 23:07:26 2009
[imud_code]
Silenus@Dead Souls Dev: well my main concern is what happens if you have a name clash?
Sun Oct 25 23:08:09 2009
[imud_code]
Wodan@Discworld: who cares? keep them strictly by the types that are in there :)
Sun Oct 25 23:09:18 2009
[imud_code]
Silenus@Dead Souls Dev: do you think the types should be ordered or not (based on the order of declaration)?
Sun Oct 25 23:10:23 2009
[imud_code]
Wodan@Discworld: yes, that'll make it easier to match types that are the actually the same
Sun Oct 25 23:11:19 2009
[imud_code]
Silenus@Dead Souls Dev: the parser = finished (for my driver) a long time ago
Sun Oct 25 23:11:32 2009
[imud_code]
Silenus@Dead Souls Dev: it's the code generation and vm stuff i am working on
Sun Oct 25 23:12:00 2009
[imud_code]
Silenus@Dead Souls Dev: the vm stuff is easier the code generation is the tricky bit
Sun Oct 25 23:12:21 2009
[imud_code]
Wodan@Discworld: can't generate code if you don't know the vm though :)
Sun Oct 25 23:13:00 2009
[imud_code]
Silenus@Dead Souls Dev: well LLVM takes care of a lot of that for me
Sun Oct 25 23:13:41 2009
[imud_code]
Silenus@Dead Souls Dev: so the runtime system becomes pretty easy to build (except for the memory management)
Sun Oct 25 23:14:41 2009
[imud_code]
Silenus@Dead Souls Dev: have about 2 weeks b4 new job starts maybe a little more we will see how far i get.
Sun Oct 25 23:15:14 2009
[imud_code]
Wodan@Discworld: don't let it mess up the start of the new job :)
Sun Oct 25 23:16:50 2009
[imud_code]
Silenus@Dead Souls Dev: just fiddling with it when i need to relax a bit :)
Sun Oct 25 23:18:09 2009
[imud_code]
Wodan@Discworld: yeah, nice and relaxing :D time to sleep, afk!
Mon Oct 26 18:40:25 2009
[imud_code]
Ideysus@ShadowMUDii: c++/java style variable declarations (wherever you want, not just before any function calls) would probably make fluffos irresistable...
Mon Oct 26 18:41:51 2009
[imud_code]
Ideysus@ShadowMUDii: I assumed porting to ld would be a pretty huge undertaking.
Mon Oct 26 18:42:33 2009
[imud_code]
Ideysus@ShadowMUDii: I think they don't support the same closure syntax, I'm pretty sure that the whole socket thing is different... I don't remember what else... but I thought there was a lot more
Mon Oct 26 18:43:04 2009
[imud_code]
Silenus@Dead Souls Dev: ld supports the (: :) stuff now as well
Mon Oct 26 18:43:35 2009
[imud_code]
Ideysus@ShadowMUDii: ... real closures? So you don't need the $() for local variables?
Mon Oct 26 18:43:39 2009
[imud_code]
Silenus@Dead Souls Dev: that might be a problem though for certain types of code since it doesnt support $()
Mon Oct 26 18:45:11 2009
[imud_code]
Ideysus@ShadowMUDii: Bah. In an ideal world I think closures would use (::) syntax and support local variables directly and bound with $()
Mon Oct 26 18:45:28 2009
[imud_code]
Ideysus@ShadowMUDii: But hey, I'm in no position to complain because I don't see myself driver hacking any time soon :-)
Mon Oct 26 18:47:59 2009
[imud_code]
Silenus@Dead Souls Dev: I considered implementing closures at one point
Mon Oct 26 18:48:32 2009
[imud_code]
Silenus@Dead Souls Dev: but i decided i would leave it for later
Mon Oct 26 18:48:58 2009
[imud_code]
Silenus@Dead Souls Dev: i think implementing closure type support requires i do custom frame management
Mon Oct 26 18:49:18 2009
[imud_code]
Silenus@Dead Souls Dev: which is quite a lot of work since LLVM comes with stack frames for free so to speak
Mon Oct 26 18:50:12 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. I don't know anything about your VM, but the $() hack doesn't need stack frames to hang around.
Mon Oct 26 18:50:36 2009
[imud_code]
Ideysus@ShadowMUDii: I wonder if true closures can be done without keeping stack frames around too...
Mon Oct 26 18:50:59 2009
[imud_code]
Silenus@Dead Souls Dev: the easy way to do it is to stick frames on the heap
Mon Oct 26 18:51:05 2009
[imud_code]
Ideysus@ShadowMUDii: I mean, could the compiler detect variables used in closures and ... yeah
Mon Oct 26 18:51:31 2009
[imud_code]
Ideysus@ShadowMUDii: I mean, I simulate closures in mudos by using a single element array
Mon Oct 26 18:54:27 2009
[imud_code]
Silenus@Dead Souls Dev: well just trying to clean up the driver code i have atm and post it somewhere
Mon Oct 26 18:59:05 2009
[imud_code]
Silenus@Dead Souls Dev: anyhow supporting declare vars anywhere with proper scoping is quite easy with LLVM. I actualy had it in my prototype.
Mon Oct 26 19:00:04 2009
[imud_code]
Silenus@Dead Souls Dev: mudos compat driver (probably minimalistic) with JIT support
Mon Oct 26 19:04:17 2009
[imud_code]
Ideysus@ShadowMUDii: Nulls in strings? I've wanted that for a while. :-)
Mon Oct 26 19:04:45 2009
[imud_code]
Ideysus@ShadowMUDii: The comma operator thing is an interesting choice...?
Mon Oct 26 19:05:14 2009
[imud_code]
Silenus@Dead Souls Dev: well i might preserve the comma operator
Mon Oct 26 19:05:26 2009
[imud_code]
Silenus@Dead Souls Dev: since it seems like no one likes tuples :D
Mon Oct 26 19:06:31 2009
[imud_code]
Silenus@Dead Souls Dev: most of the language specifics are easy to hack in or out
Mon Oct 26 19:07:51 2009
[imud_code]
Silenus@Dead Souls Dev: was thinking of merging the buffer/string types and having pascal strings
Mon Oct 26 19:08:23 2009
[imud_code]
Silenus@Dead Souls Dev: internally it will be length + char array but no idea what the interface will be
Mon Oct 26 19:20:50 2009
[imud_code]
Silenus@Dead Souls Dev: not sure which project page you are looking at but a lot of idea are probably out of sync with what i am doing now. I suspect the more aggressive semantic changes are out.
Mon Oct 26 19:22:26 2009
[imud_code]
Silenus@Dead Souls Dev: LPC on LLVM is probably where i will stick the code
Mon Oct 26 19:22:43 2009
[imud_code]
Silenus@Dead Souls Dev: but it might take me months to finish it
Mon Oct 26 19:23:09 2009
[imud_code]
Silenus@Dead Souls Dev: since my free time will be disappearing with the new job.
Mon Oct 26 19:27:02 2009
[imud_code]
Kalinash@Fire and Ice: you should be, at the same time, earning 2 PHDs as well as writing a mud driver, mudlib and running a game
Mon Oct 26 19:36:04 2009
[imud_code]
Detah@Dead Souls Dev: "whatcha mean you only got one job?" "And don't give me any o' dat 'I'm only three years old crap.'"
Mon Oct 26 23:28:58 2009
[imud_code]
Ideysus@ShadowMUDii: Anybody got any interesting ideas on object referencing? Old nightmare uses something like set_id("hat", "red hat") and so on. The driver parser uses a system of nouns and adjectives (I forget, perhaps something like set_nouns("hat") set_adjectives("red") ). Are there any better alternatives?
Sun Nov 1 16:23:13 2009
[imud_code]
Ideysus@ShadowMUDii: Have any LPCers ever used something like markdown?
Sun Nov 1 16:24:06 2009
[imud_code]
Ideysus@ShadowMUDii: I'm thinking about help files and I wonder if something like markdown or perhaps dokuwiki syntax might be a nice way to handle such text.
Sun Nov 1 16:25:11 2009
[imud_code]
Ideysus@ShadowMUDii: It would make transformation to html easier and would allow for some nicer formatting on the mud. Perhaps links could be presented in something like mxp or whatever too.
Sun Nov 1 16:38:44 2009
[imud_code]
Kejope@Cookie Dough: I do like the idea. When I researched some possibilities a while back, I concluded that they all still have one flaw I didn't feel like dealing with: They all have some -- different -- method for marking, which would need to be learned (yet another 'language' to learn) in order to be useful. I may not mind dealing with that, but don't want to fight with colaborators not using it. In the end, I concluded that, unless I am working in some giant project, sanity requires me to not invest the time in that direction. I mostly use "line position" to indicate meaning, sort of like the format of a man page: First line is always the name and a brief description, Second line is always the synopsis, or syntax of the function, Next lines are always the individual parameters and descriptions (one per line, no matter how long the line length), then a RETURNS line, after which the next lines describe any useful return values, then a NOTES line, followed by any other information that seems useful.
Sun Nov 1 16:41:33 2009
[imud_code]
Ideysus@ShadowMUDii: Yeah, I see that differing syntaxes is an issue.
Sun Nov 1 16:42:48 2009
[imud_code]
Kejope@Cookie Dough: To me, my method seems easily parsable, when and if I ever have a need to HTML-ize it. Any formatting could be applied by an interpreter, so that, for example, the parameter name (i.e., --version) is recognizable at the start of the line, and could be bolded easily.
Sun Nov 1 16:42:50 2009
[imud_code]
Ideysus@ShadowMUDii: I'm going to think about either an lpc based dokuwiki/markdown syntax parser though
Sun Nov 1 16:44:02 2009
[imud_code]
Kejope@Cookie Dough: .. and I would be happy to consider it when it is ready, for certain definitions of "ready." :)
Sun Nov 1 16:46:53 2009
[imud_code]
Ideysus@ShadowMUDii: Interesting. nroff seems to be more about formatting plain text, rather than a way of identifying semantics of text.
Sun Nov 1 16:52:48 2009
[imud_code]
Ideysus@ShadowMUDii: Maybe I'm wrong... I see htmlised nroff output...
Sun Nov 1 16:53:39 2009
[imud_code]
Ideysus@ShadowMUDii: I'm tending away from the thought of nroff though. It looks more complicated, though my initial perception may be incorrect.
Sun Nov 1 16:54:48 2009
[imud_code]
Kejope@Cookie Dough: See? There's that fighting I was talking about. hehe
Sun Nov 1 16:56:16 2009
[imud_code]
Ideysus@ShadowMUDii: Thats my motivation for something like markdown. It looks reeeeally easy. And in the worst case if somebody just writes plain text I assume they'll just get plain text output.
Sun Nov 1 16:59:01 2009
[imud_code]
Kejope@Cookie Dough: I agree. Markdown was going to be it for me if anything were to be it. :)
Sun Nov 1 17:04:31 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. I'm now tending towards a markdown subset. Markdown seems html specific.
Thu Nov 5 14:42:38 2009
[imud_code]
Silenus@Dead Souls Dev: how is javascript different from most OO languages?
Thu Nov 5 15:05:03 2009
[imud_code]
Kalinash@Fire and Ice: but you can still do data hiding and inheritance and polymorphism so it can be used as an OO language
Thu Nov 5 15:05:12 2009
[imud_code]
Kalinash@Fire and Ice: i can send you some good links when i get to the office
Go to the top | Channel Index
