Announcement

**Locutus** · October 28, 2003, 22:31

Yacc (.y files)
[21:26] <Solver> Performing Custom Build Step on .\ctpdb.y
[21:26] <Solver> wtf are .y files and why are they built?
[21:26] <Solver> it has code, but aint recognized as a code file
[21:26] <Solver> oh well, fun has started
[21:26] <Solver> loc one thing you might like
[21:27] <Solver> it builds to civctp.exe not ctp2.exe

[22:15] <SprudL> Solver: is it possible those .y files contain yacc stuff?
[22:15] <Solver> what's that

?
[22:15] <Solver> They contain c-like code, but surrounded with % signs, for instance
[22:16] <SprudL> don't know if it exists on windows but I've heard about it on linux ... yacc generates code
[22:16] <SprudL> let me see if I can find a manual
[22:16] <Solver> hardly... they contain normal code
[22:17] <Locutus> yacc = yet another compiler compiler. used for creating your own compiler. might have been used for SLIC....
[22:17] <SprudL> http://dinosaur.compilertools.net/#yacc
[22:17] <Solver> nah, it's in the UI section
[22:18] <Locutus> sounds unlikely then, though I'd have to see it myself. shame on you for not knowing yacc, Solv

[22:18] <Solver> shut up, loc

[23:18] <Locutus> btw, I noticed a byacc file in the bin folder, so the .y files might refer to yacc files after all, SprudL (if you're still here)
[23:19] <SprudL> indeed
[23:19] <SprudL> but there's allso an ldlyacc.h or something
[23:19] <SprudL> that more or less gave it away
[23:20] <J_Bytheway> I could have sworn I saw bison somewhere, but perhaps I imagined it...
[23:20] <Locutus> ah. then the LDL files form their own little language (like SLIC) and are compiled by a yacc-generated compiler...
[23:20] * MRex has joined #apolyton
[23:20] <Locutus> I think I saw that name as well... hmmm...
[23:21] <MRex> hi

[23:21] <SprudL> so there could a language to make the UI then?
[23:21] <Locutus> maybe they used yacc for LDL and bison for SLIC?

[23:21] <SprudL> could *be* a language... djz.. still can't type
[23:21] <Locutus> well, the LDL files sure look like a scripting language. But noone has ever been able to make much sense of them.
[23:22] <Locutus> and yes, the LDL files control the UI
[23:23] <Locutus> open the LDL files in a text editor and see for yourself...
[23:23] <SprudL> there's a slic.y file
[23:23] <Locutus> hmm...
[23:23] <SprudL> yacc for slic too then?
[23:23] <Locutus> the plot thickens. would make sense. but what's bison doing there then? or maybe John and I were just dreaming...
[23:24] <J_Bytheway> I remember when I first saw ldl files in CTP1 and I was most dissapointed because I had no idea how they worked, and I was most impressed that anyone had modded them - it was over a year before I thought to open one in a text editor.
[23:24] <J_Bytheway> I think perhaps I saw the name "bison" appear fleetingly during the installation...
[23:25] <Locutus> aha... bison is mentioned in the readme. according to that, there should be a bison file in the bin folder but there isn't...
[23:25] <Zephyr> its in the readme IIRC
[23:25] <SprudL> can't find bison at first glance
[23:25] <Locutus> doh! byacc!
[23:25] <Zephyr> dammit, crossspeak.
[23:26] <J_Bytheway> Yep, it was in the readme
[23:26] <J_Bytheway> That explains it.
[23:26] <Zephyr> byacc == bison?
[23:26] <G_Slacker> I'm not sure of this, but I think bison and yacc are two different things that perform the same function.
[23:26] <Locutus> aha. from gnu.org: Bison is upward compatible with Yacc: all properly-written Yacc grammars ought to work with Bison with no change.
[23:26] <SprudL> yes
[23:27] <SprudL> somehow they are related
[23:27] * MRex has left #apolyton
[23:27] <SprudL> hmz.. locutus found out already

[23:27] <Locutus> so they used Bison, not Yacc -- and bison is basically yacc++

wasn't aware of that -- thought they were completely seperate things
[23:28] <Locutus> so that's a step forward. we know the .y files are bison files
[23:28] <J_Bytheway> So is byacc.exe actually bison?
[23:29] <Locutus> yes. at least I think so.
[23:29] <SprudL> maybe it can handle both?
[23:29] <Locutus> probably. seeing how bison is yacc compatible...
[23:31] <Locutus> okay, did a search: the .y files are ctpdb.y, slic.y, sliccmd.y, fli.y and ldl.y
[23:31] <Locutus> so they used bison/yacc for their internal database, slic, AI and UI...
[23:31] <Zephyr> two slics?
[23:32] <G_Slacker> I think sliccmd.y is a list on slic commands. Based purely on the file name.
[23:33] <Locutus> yeah, not sure what that means either. but in antlr (the compiler compiler I'm most experienced with) you can just string text files together and it'll treat it as a single file -- just like SLIC. same might be the case here. the slic.y file *is* rather large, so splitting it up makes sense.
[23:35] <Locutus> oh yeah... just looking at slic.y now. I don't know yacc very well, but this DEFINITELY looks like a compiler compiler...
[23:35] <J_Bytheway> Indeed, the lexer is in slic.l.
[23:36] <Zephyr> ooh, a comment
[23:36] <SprudL> apparently byacc stands for berkeley yacc
[23:36] <SprudL> I thought is was bison/yacc
[23:37] <Locutus> hmmm... strange. either way, it's yacc-ish. I'm sure we can deal with that

[23:37] <SprudL>

[23:37] <J_Bytheway> I've been playing around with compiler constructors a lot recently, with this partly in mind.
[23:37] <Locutus> not that we need to really. most of the stuff we'll want to change is probably in the implementation, not in the definition of the various languages used
[23:38] <J_Bytheway> I note also that slic.y has some debugging stuff under #ifdef directives. Might be handy.
[23:38] <Locutus> damn, I knew I should have finished that class

[23:38] <Locutus> yeah, I noticed.
[23:39] <J_Bytheway> I have half a mind to try to write an entirely new sscripting language from scratch... Perhaps that's a little overambitious, but it might be easier than twisting SLIC too far.
[23:39] <J_Bytheway> And it makes backwards-compatibility easier too.
[23:40] <Locutus> in the long run that might be a good idea, in which case we'll have plenty of time to figure out what version of yacc it is

in the short run we'll want to focus on fixing the existing language.
[23:41] <SprudL> I think getting familiar with the source tree will be hard enough at first

Compiling
[00:29] <J_Bytheway> I can't seem so compile anything other than crater.dll...
[00:29] <J_Bytheway> I can't remember how this IDE works any more...
[00:29] <Locutus> hmm, bummer. are you getting errors on other files?
[00:33] <J_Bytheway> I wasn't getting errors, I just couldn't figure out how to instruct the IDE to compile them. Whatever I did it compiled crater.dll and nothing else.
[00:33] <J_Bytheway> I have now managed plasma2.dll too, though

[00:34] <Locutus> ah, ok. that's good news then -- at least the source is not buggy

[00:34] <Locutus> w00t, progress

[00:36] <J_Bytheway> OK, now I'm having problems because it doesn't know how to use flex and byacc - I need to set that environment variable...
[00:36] <Locutus> solver was running into that as well. then he quit. but then we didn't know about yacc yet, so hopefully you can it things a step further now

[00:43] <J_Bytheway> I can't make it recognize byacc...
[00:43] <J_Bytheway> I'm getting "'\byacc' is not recognized as an internal or external command"
[00:44] <J_Bytheway> Hmm...
[00:44] <J_Bytheway> I feel that the backslash before byacc can't be helping...
[00:45] <Locutus> probably not...
[00:45] <J_Bytheway> fault.dll works, though. geometric is failing to recognise the environment variable which I've added. Perhaps a restart is in order.
[00:47] <Locutus> probably. not sure about XP but it would be under older OSes...
[00:47] <J_Bytheway> Anyone know how to edit custom build settings under VS6?
[00:47] <Locutus> nope
[00:51] <J_Bytheway> MSDN is little help on the subject either. I'll try rebooting.

[00:59] <J_Bytheway> OK, rebooting fixed the environment variable problem, now there's just byacc to contend with.
[00:59] <Locutus> good, what problems are you having with byacc?
[01:01] <J_Bytheway> Ah ha! I have found the custom build settings

[01:01] <Locutus> good...
[01:03] <J_Bytheway> But now I see that the error message has in fact changed, now it just says "Performing Custom Build Step on ..\ui\ldl\ldl.y"
[01:03] <J_Bytheway> And then "The system cannot find the path specified."
[01:03] <J_Bytheway> Which is most unhelpful.
[01:03] <Locutus> very helpful indeed...
[01:07] <J_Bytheway> I'll try compiling it by hand...
[01:09] <J_Bytheway> That worked

[01:09] <J_Bytheway> Now another even more cryptic error message...
[01:09] <Locutus> it did?

[01:09] <Locutus>

[01:09] <Locutus> being?
[01:11] <J_Bytheway> Well, if you look in ctp2_code\gs\newdb\ you will find several .cdb files
[01:12] <Locutus> I do.
[01:12] <J_Bytheway> These are supposed to be custom-built by something called ctpdb.exe, hich we don't seem to have.
[01:12] <J_Bytheway> *which
[01:12] <Locutus> interesting... maybe we have to build it ourselves?
[01:13] <J_Bytheway> Alas, it's not immediately obvious what they're supposed to build to...
[01:14] <J_Bytheway> Aha! The other project that wasn't compiling seems to be the source of this executable. Don't give up hope yet

[01:14] <Locutus> told you

[01:15] <Locutus> first impression is that the cdb files are the scripting language for the internal database of the game, the one defined by ctpdb.y
[01:17] <J_Bytheway> Yes, sorry, I didn't understand when you said that

. That compilation was also stuck on a yacc file (ctpdb.y) so I've done that one manually...
[01:18] <J_Bytheway> And now the lexer file is stumping it...
[01:19] <Locutus> hmm, well whoever said progress was slow sure as hell was talking about us

[01:19] <Locutus> but progress is progres nonetheless...
[01:20] <J_Bytheway> Got it! Now only ctp2.exe remains to be dealt with

[01:20] <Locutus>

[01:21] <Locutus> incredible, at this pace you may actually get it working before 6 AM

*knocks on wood*
[01:21] <J_Bytheway> Now I get "discombobulating sliccmd.y" - does that sound like st_swithin to you?

[01:21] <Locutus>

[01:23] <J_Bytheway> Well, I might at least reach a Miles sound system error by 6AM...
[01:23] <J_Bytheway> That is likely to stop me in my tracks

[01:23] <Locutus> well, we should only have to outcomment it, if we're lucky...
[01:24] <J_Bytheway> Now "burgling slic.y" - Someone was having fun with these.
[01:24] <Locutus> hmm, that's my lemur!

[01:25] <J_Bytheway> Well well, SLIC syntax has a shift/reduce conflict. tut tut.
[01:25] <Locutus> uhuh...
[01:26] <Locutus> aha, I just found the 'makefile' for the project. you were right: according to the comments, you should be able to make a regular makefile out of it with Export Makefile...
[01:27] <J_Bytheway> The next one is "pummeling DBLexer.l"
[01:27] <J_Bytheway> Do you mean that there is a makefile there already?
[01:27] <Locutus>

[01:28] <Locutus> no, not really, an Microsoft Developer Studio Project File
[01:28] <Locutus> but it's the VB equivilant of a makefile

[01:29] <J_Bytheway> It would probably be faster if I figured out why all these lexings and parsings weren't working rather than doing them all by hand. It's quite strange, because if I copy the commands they should be executing onto the command line it works...
[01:30] <Locutus> that is very strange indeed.
[01:32] <J_Bytheway> I guess perhaps it is executing them in the wrong directory...
[01:33] <Locutus> well, if the problem is still 'path not found', then I would guess so, indeed.
[01:33] <J_Bytheway> "snorkeling slic.l" - this is getting a little silly...
[01:33] <Locutus>

well, I'm still thoroughly enjoying it...

[01:35] <Pete_T> At this level, I suspect JBR (Joe)
[01:35] <Locutus> but judging from the project file, that should be the last of them.
[01:35] <J_Bytheway> "combobulating sliccmd.l"
[01:35] <J_Bytheway> Very different from the prior discombobulation, I'm sure.
[01:35] <Locutus> oh, I thought you already did that one... but this one IS the last one then

[01:36] <Locutus> well, it WAS him who designed SLIC...

**Locutus** · October 28, 2003, 22:36

[01:38] <J_Bytheway> Right, that's the last of flex and byacc - we're now onto the real compilation for ctp2.exe, and I think that I've hit the miles sound system errors - I have unfound mss.h
[01:39] <Locutus> aha, that is indeed a mls error...
[01:39] <J_Bytheway> Perhaps surprisingly (for me, at least) most of the files are compiling without problems.
[01:40] <J_Bytheway> And they go on and on and on....
[01:40] <Locutus> there are 4 references to this mss.h file in the code. just uncommenting them will hopefully do the trick...
[01:40] <J_Bytheway> and on and on...
[01:40] <Locutus> and two of those 4 are in the readme

[01:41] <J_Bytheway> The main one is in civsound.h, but unfortunately that one is referenced by many others.
[01:42] <J_Bytheway> And I've just seen some other errors too...
[01:42] <Locutus> the other is in the project file, so hopefully that one shouldn't give any problems
[01:42] <J_Bytheway> It's still going. Now I begin to see how much ~2million lines of code really is...
[01:43] <Locutus> LOL...
[01:43] <J_Bytheway> I've never written anything that took more than a few seconds to compile...
[01:43] <Locutus> can't you abort?
[01:43] <Locutus> well, that's not much... I can beat that, but I'm not sure what my 'record' is...
[01:44] <J_Bytheway> I could abort, but I don't want to. It has to be done sometime, and it won't need to do it again.
[01:44] <J_Bytheway> and on and on....
[01:44] <Locutus> well, if there's all sorts of errors, you'll have to compile again anyway, won't you?
[01:45] <Pete_T> How long has this been taking? I've seena book called 'Game Programming in Linux' where the author had access to the CTP1 code and said that it took two hours to compile
[01:45] <J_Bytheway> But there aren't - there are very few errors. That's what surprised me

.
[01:45] <J_Bytheway> It's been going for less than 5 min, I guess. Perhaps I should be a little more patient

.
[01:46] <Locutus> well, not being able to find mss.h is an error, or is that just a warning?
[01:46] <J_Bytheway> It's an error, but it affects only about 1 file in 50.
[01:46] <J_Bytheway> The rest compile OK.
[01:46] <Locutus> I'm still looking for a compiler that has one of those neat counters that keeps track of how much % of the work was done, but for some reason I can't seem to find any

[01:47] <J_Bytheway> If I ever write a compiler, I'll try to keep that in mind

[01:47] <Locutus> I would owe you bigtime

[01:47] <J_Bytheway> But I should point out that it's the IDE that should to the counting, not the compiler.
[01:48] <Locutus> well, there's only about 2632 files, so that shouldn't take so long

[01:48] <J_Bytheway> Done - 101 errors, 5 warnings.
[01:49] <Locutus> hey, that's not bad at all... for 1,8 million lines of code that's actually very, very good

[01:49] <Locutus> and it didn't take too long either.
[01:50] <J_Bytheway> So shall I start commenting stuff out ans see what happens?
[01:50] <Locutus> so what kind of errors? aside from the obvious mss.h missing?
[01:50] <Locutus> yeah, sounds good.
[01:51] <Pete_T> For sure. I was hoping the CTP2 code would be a bit cleaner.
[01:51] <J_Bytheway> Most of those were mss.h - some were undefined things which I didn't really catch...
[01:51] <Locutus> surely you have a log?
[01:52] <Locutus> but just outcomment and see what happens. it's always a good feeling to see the number of errors drop from 100 to 20, even if nothing really changed

[01:52] <J_Bytheway> Well, the log is not easily accessible now that I've started another build...
[01:53] <Locutus> ah, ok. same file, without mss.h, I assume?
[01:53] <J_Bytheway> With that outcommenting we're now getting approx 4 errors for each one we had before

[01:53] <J_Bytheway> Due to all the references to things in mss.h...
[01:54] <J_Bytheway> That's 517 errors, 57 warnings

[01:55] <Pete_T> That makes sense. Can you save the log?
[01:55] <J_Bytheway> I can try cutting out the whole civsound class, which will cascade a whole new set of errors...
[01:55] <J_Bytheway> I can indeed save the log. Probably a good plan

[01:56] <Locutus> fun

[01:56] * ACS_MarkG has quit IRC (Quit: Leaving)
[01:56] <Locutus> chopping civsound sounds good, at least as a start. we'll have to rewrite it anyway.
[01:57] <J_Bytheway> My usual programming style fits well with this. I usually do something along the lines of "fix one error, recompile"^N
[01:58] <Locutus> don't we all?

[02:01] <J_Bytheway> Well, I mean I do it in a slightly more extreme fashion than normal. Some people prefer to stop with versions that actually work now and again. I once wrote about 15000 lines and I think my compilation was successful no more than 5 times during that development. Then I got bored and gave up

.
[02:01] <J_Bytheway> OK, now we have 923 errors, 16 warnings

[02:02] <Locutus> lol, well it helps to cut things in pieces...
[02:02] <Locutus> this is not getting any better...
[02:02] * Panzer32-away is now known as Panzer32
[02:02] <Locutus> lots of the same (kind of) errors or different ones?
[02:03] <J_Bytheway> Now they're mostly errors on the lines which say "CivSound", which is now unrecognized.
[02:04] <J_Bytheway> Perhaps we would be better off leaving CivSound in, but nullifying its routines in some fashion...
[02:05] <Locutus> yes, overwrite them with stubs, sounds like a good plan.
[02:06] <J_Bytheway> But I think I'm going to have to go to bed. Unfortunately I have a lecture in 8 hours. I should be back at work tomorrow afternoon, unless Solver beats me to it

Full chat transcript in text file below (too big to post), but if you read all of the above, there's not much news in there, mostly noise.

Attached Files

ctp2sourcechatparty.txt (55.2 KB, 256 views)

**Solver** · October 29, 2003, 08:31

Great work Wouter.

Oh, and a big thumbs up to those who have already worked on getting things as far as they are! I admittedly didn't expect this much within hours after the code release.

**J Bytheway** · October 29, 2003, 10:07

Here are a couple of error logs for the terminally curious (both in one file). The first is what happens if I comment out just the #include "mss.h", and the second if I also comment out the whole CivSound class.

Attached Files

error_logs.txt (191.4 KB, 329 views)

**Locutus** · October 29, 2003, 10:30

Interesting, John

(Yes, I'm terminally curious

)

Announcement

Release Party Chat Transcript (lots of useful info)

Release Party Chat Transcript (lots of useful info)

Comment

Comment

Comment

Comment

Comment