00:00:51 ixtli: Nice. :) 00:05:12 ixtli: Luck, I think SDL is kinda essential at the minute :) 00:05:16 ixtli: (You mean tiless on iphone?) 00:05:27 due: Yeah. 00:05:46 Shouldn't be that hard. 00:06:14 In fact, there is less setup to be done knowing that I can count on iPhone OS to create my environment for me. 00:06:15 !tell Enne Why did you name it tilegen.elf? (c.f. b25208caa) 00:06:15 sorear: OK, I'll let Enne know. 00:06:24 lol 00:06:31 -!- DrPraetor has joined ##crawl-dev 00:06:53 due: It wont be essential when I'm done with it, though for Enne's sake I hope to make SDL's use in crawl more modularized. 00:07:09 Right now it's inside everything with #ifdef USE_TILE 00:07:13 Which works, but ... you know. 00:07:16 It aint pretty. 00:07:41 -!- DrPraetorious has quit [Ping timeout: 246 seconds] 00:07:47 I wish you success 00:08:03 I'd really like to have tiles a bit more modularized 00:08:42 Well, I'll tell you that there is probably going to be little modification to the OpenGL code to get it running on OpenGL ES 2.0 on the ipad. 00:08:46 But. 00:08:56 SDL has to go entirely, so I'll see if I can't make a more elegant solution. 00:09:34 I could use that factory design pattern to make a UI factory. 00:09:49 So we could keep all the #ifdef USE_TILE hooks in the Crawl code. 00:10:05 Actually what am I saying, those hooks are just to update state. 00:10:14 Hmm. I need to think about this a little more. 00:11:27 sorear: What specifically do you want to see? 00:11:48 sorear: What bothers you at this point? I have yet to really get in to the drawing code. 00:13:31 ixtli: I want to get rid of the mandatory SDL/OpenGL dependancy 00:13:46 Yeah. 00:14:06 factory design pattern? isn't that something you use in Java to fake global variables? But this is C++, we have real globals 00:14:23 sorear: Nah, it's just a class design pattern. 00:14:51 I think that's what I mean XD Look man, I took that class a while ago :P 00:14:53 http://en.wikipedia.org/wiki/Factory_method_pattern 00:15:27 Basically what I mean is that you have a generic class called UIManager or something, and when you instantiate it you tell it that you want to use SDL 00:16:02 Then if you want a window, you say UIManager.makeWindow() and it sorts everything out behind the scenes. 00:16:10 sorear: singleton is the one that fakes global variables 00:16:16 learn your design patterns d: 00:16:20 Then you can add your specific win32 or aqua or whatever. 00:16:47 ixtli: You should really talk to Enne Walker 00:16:53 due: I have been. 00:16:56 ah, good 00:16:58 :) 00:17:08 Enne's mind astonishes me. 00:17:16 Subvaults code is just so beautiful but I still do not understand it. 00:17:17 She's knows OpenGL and is thus praised. 00:17:30 s/\'s// 00:19:34 due: Where is this? 00:21:47 ah, so factory is the one that fakes functors. :p 00:23:31 I dunno. It doesn't matter if you don't know what these things are called. 00:23:44 I've become a bit annoyed at that assertion in academia and corporate world. 00:24:02 As long as you can understand them and make them pretty once you do, you're perfectly fine. 00:24:54 I'm trying to insert line breaks in a regexp search and replace, but when I put in \n, vim displays this blue ^@ instead of a line break. 00:24:59 night all 00:25:34 -!- Textmode has quit [Quit: Leaving] 00:25:43 Keskitalo: try \r 00:26:28 Vim stores your file as a char**, the nul handling is trippy 00:28:15 (i wasn't being serious about that 'learn your design patterns' comment, by the way) 00:28:56 \r worked! Thanks! I thought I already tried that, but apparently I failed to do so. 00:29:25 I'm trying to learn to use tools like regexp instead of manual search and replace. :) 00:29:30 ^V before the literal key works too 00:29:36 that's anywhere in vim 00:32:18 Hah! 00:32:18 char internal_buf[500]; // if your terminal is wider than this, too bad 00:32:47 doy: Assumed so :) It's just something that bothers, in general. 00:32:50 ixtli: Let's not look at the make-stringf code. 00:33:10 due: The teleport-to -vaults - they lack the tiles, right? And they'd need them? 00:34:06 ixtli: was that re-added after the va_copy fiasco? 00:34:17 sorear: I've been here for two days XD 00:34:30 ixtli: ok, what file is that statment in 00:34:46 class mpr_stream_buf in message.h 00:35:19 Another thing I might try my hand at doing is grepping the entire source bace for comments that end in "?" and putting a TODO: in front of the 00:35:21 them* 00:42:46 -!- Spads has quit [Ping timeout: 256 seconds] 00:48:41 Wow. GCC really saturates my processors. 00:49:36 Do troves need milestones? Bazaars don't have them. 00:50:03 Keskitalo: I'd give bazaars and labyrinths miletones. 00:51:09 I can do them for bazaars. 00:53:56 -!- syllogism has joined ##crawl-dev 01:07:08 sorear: Can we talk a bit about DEFINEs for modularizing UI/GL code? 01:07:27 aaa aaa aaa misuse of can aaa 01:07:43 MAY?! 01:07:57 see also: irc rule #1, don't ask to ask 01:08:04 just ask already 01:08:07 lulz 01:08:18 First of all! I've been in Japan for 5 months, cut me some slack. 01:08:22 re: english 01:08:34 i'd kick you if it wasn't so much of a bother to go op myself d: 01:08:55 I cut people slack when they don't ask to ask 01:09:08 Just making sure you've got time, jeeeeeeeze. 01:09:09 Anyway. 01:09:21 I don't think it's right to cut out USE_TILE 01:09:36 why not? 01:10:08 we should at least be using conditional linking 01:10:16 ui-tiles.o or ui-curses.o 01:10:17 Yes. 01:10:19 However 01:10:21 link ONE into the binary 01:10:27 because it appears that 95% of the time when #ifdef USE_TILE is in the crawl source 01:10:28 and all dependant libraries 01:10:35 in order to update state. 01:11:09 that sentence was badly structured. 01:11:26 The code seems to check USE_TILE a lot before setting state in the GUI 01:11:34 which is platform/renderer agnostic. 01:12:24 hey, can I ask a question? 01:12:34 will people weep if there is no play-testing post? 01:12:51 What was it supposed to be about? 01:13:06 It's the big list summary. 01:13:12 It's... taking me a lot longer to do than I expected. 01:13:34 yeah, people have been ridiculously active these past bunch of days 01:13:46 moin 01:13:46 cherry-picking is taking an annoying amount of time too(: 01:13:51 Zaba: Hey :) 01:13:51 Meh, I get an empty "Lua error:" with the epilogued milestone functions. 01:13:58 Keskitalo: really? 01:14:05 I was going to ask about whether or not crawl-dev is always this active. 01:14:07 bugger me 01:14:18 doy: 264 commits in the past week 01:14:27 ##crawl-dev has gotten quite active lately 01:14:34 due: don't feel pressured or anything, whatever you can do in your free time(: 01:14:34 due: Maybe it doesn't like passing the e/_G argument? The pan lord messages worked fine, but they were just crawl.mprs. 01:14:54 ##crawl-dev hasn't been all that much more active, really 01:15:00 Keskitalo: e/_G arguments probably won't work, now that I think about it. 01:15:05 it's always been pretty busy 01:15:09 Keskitalo: But the blank error messages is something else 01:15:11 but the commit rate has been extra high 01:15:12 sorear: So, the issue here is that there are like 300 #ifdef USE_TILE in the crawl source 01:15:15 due: Okay. 01:15:35 * doy points to file sizes here http://tozt.net/crawl/ 01:15:39 !tell greensnark Can you take a quick look at the epilogue code? I think I've messed up with the error reporting somewhere along the line. 01:15:39 due: OK, I'll let greensnark know. 01:15:50 Keskitalo: Solution: summon el snarko. 01:16:02 weekends are also always quite a bit busier 01:16:05 ixtli: They all need to die. 01:16:15 sorear: Sounds good to me. 01:16:53 That's problematic, though. 01:16:59 But what's going to have to happen in the long run is a lot of old code in crawl is going to have to be factored out into a library that treats text drawing in the same way as gl drawing 01:17:03 yes, it is. 01:17:15 ixtli: That's not even half the trouble. 01:17:22 As you said, ui-curses 01:17:22 ixtli: Go look at the code that's IFDEF'd out in mapdef.cc. 01:17:36 due: I have a feeling I don't want to :( 01:18:03 ixtli: how about we not worry about it this month, we have a 0.6 to ship 01:18:10 Most of it is because the tile headers and relevant functions aren't compiled for normal builds, therefore meaning that "TILE: " does something different on a tiles build and a non-tiles build. 01:18:14 sorear: 0.6 is already branched 01:18:26 you guys can do whatever you want on master 01:18:27 ??firefox 01:18:27 I don't have a page labeled firefox in my learndb. 01:18:36 what's the name of hte card with the buggy things 01:18:40 ??foxfire 01:18:40 I don't have a page labeled foxfire in my learndb. 01:18:41 vapours and stuff 01:18:45 foxfire! thank you. 01:19:14 anyway, part of the issue is that tiles and non-tiles provide different amounts of information 01:19:23 due: Eww. 01:19:26 and have different requirements in how things are drawn 01:19:40 such as tile variants for things that are logically the same glyph 01:19:41 er 01:19:44 doy: Eww 01:20:31 saying "all USE_TILE stuff should die" is great, but it would be a pretty serious reorganization of a huge part of the code 01:20:49 wow slings do a lot of damage 01:20:55 doy: Yeah. 01:21:03 syllogism: This is not new. 01:21:25 doy: However, modularizing drawing/wm code is something a lot of people seem to want to do. 01:21:35 steel bullets are though 01:21:38 ixtli: well, yes 01:21:56 doy: And I guess this is the conclusion, that it's going to require a lot reorganization of the old crawl stuff. 01:22:18 ixtli: but modularizing the tiles drawing code so that it can use different backends and modularizing the drawing code in general so that we can use the same drawing code for console and tiles are two very different things 01:22:24 involving two very different amounts of work 01:22:27 (: 01:22:28 Indeed. 01:22:36 However, this leads back to my initial question: 01:22:47 Assuming I'm going to leave USE_TILE define in there. 01:22:53 How should I structure my defines? 01:23:03 what is your goal? 01:23:15 doy++ 01:23:25 what stuff are you going to need to do outside of the tiles-specific code? 01:23:33 To remove dependency on SDL. 01:23:52 do we make explicit sdl calls outside of the tile*.cc files? 01:23:56 Nope. 01:24:05 However SDL_main is included in main.cc 01:24:17 well, that should be split out then, probably 01:24:31 I was thinking of implementing a class called UIManager 01:24:46 That could be initialized with a flag telling it which libs you want to use 01:25:37 Which would then spit out a class you could message for UI things like delay() and refresh(), so you wouldn't have to ever care if it's SDL or UIKit or win32 or whatever 01:26:03 well, that's one way to do it 01:26:28 it depends on if you want to be able to use multiple backends from the same compiled copy, i guess 01:26:31 The thing is, it'll still require preproc flags like USE_SDL or USE_UIKIT 01:26:46 well, no 01:26:48 it shouldn't 01:27:02 because the implementations of the sdl-specific bits should all be in their own file 01:27:15 and you just care about whether that file was linked in 01:27:22 i mean, it might require that in like one spot 01:27:33 but i don't see that as being something that needs extra worrying about 01:27:44 if you're going to go as insano impractical as a UIManager, why not go the whole way and dlopen sdl at runtime? 01:27:57 then we can compile all the backends at once 01:28:01 well, i don't think it's necessarily impractical d: 01:28:10 lol this is why I asked, I didn't think its impractical 01:28:13 it just depends on what our goals are 01:29:36 in any case, back to the original question 01:29:53 why do you think compile flags for specific backends are an issue? 01:30:21 ??sling 01:30:21 slings[1/2]: Damage rating: 0 Accuraccy rating: +2 Base attack delay: 110%. Better than throwing rocks at a jelly, but only until you wield a cursed one. Can launch {stones} and {sling bullets}. 01:30:33 so delay 5 at high skill 01:31:05 syllogism: assuming melee and ranged weapons use the same speed formula... 01:31:07 is that 2 with haste 01:31:13 (i don't know if that's true or not) 01:31:14 doy: do you know they dont? 01:31:29 syllogism: the melee and ranged speed formulas have nothing to do with one another 01:31:31 no idea, but from what i know of the combat code, i wouldn't say it's a safe assumption(: 01:31:48 the ranged speed formula is from 4.1, the melee speed formula is from 4.0 01:31:51 looking ad vandal I wouldnt be surprised if it was close 01:32:06 it's so fast because of his stats 01:32:09 not his skill 01:32:17 slings are what are fast 01:32:19 oh that's weird 01:32:26 fastest ranged weapon in the game 01:32:35 fsim choked when I tried testing 01:32:39 it's stuck at 0 01:32:48 fsim is probably melee-only 01:33:39 fsim theoretically supports all forms of combat 01:33:49 fsim is also one of the most unloved parts of Crawl 01:33:55 second to the DOS port :p 01:34:02 I love it :P 01:34:11 syllogism: you're not a coder 01:34:11 (: 01:35:03 doy: So, tilereg.cc and view.cc all include tilesdl.h directly. Also, the class TilesFramework refers to SDL types directly in it's definition. 01:35:22 ixtli: okay 01:35:35 sorear: what is the ranged speed formula? 01:36:18 doy: something like base - (weighted stat * skill) in general, speed is *2/3 01:37:03 launcher_final_speed 01:37:18 !source launcher_final_speed 01:37:19 Lines pasted to http://pastebin.com/m7a904b84 01:37:43 doy: So what is the best way to abstract those two from interacting directly with tilesdl.h? I'm thinking a more generic version of TilesFramework 01:37:47 throwing is always speed 10 01:37:54 so much for large rock power, huh 01:38:12 doy: However I don't know what the best way to detect what libraries to use would be. (If preproc defines aren't the way to go) 01:39:25 ixtli: again, why are you saying that preprocessor defines aren't the way to go? 01:39:48 Heh, maybe I misread something up there. I thought they were being discouraged, my bad. 01:40:10 ixtli: preprocessor defines for this purpose would end up being used in like one or two places 01:40:12 Then the question becomes how should I structure them, assuming that we need to leave USE_TILE in there for the time being. 01:40:16 if you structured it properly 01:40:26 i can't imagine how that would be an issue 01:41:01 So I'll make new ones for USE_SDL and USE_UIKIT, for the time being? 01:41:07 yes 01:41:27 Cool, sorry to go around and around with this for so long. 01:42:58 i wonder if the ev bonus from dex should at least be stepped down 01:43:13 i mean, 63 ev is pretty ridiculous 01:43:39 I wonder if the dex bonus from ponderousness should at least be stepped down 01:43:48 i mean, d8 dex is pretty ridiculous 01:44:12 i wonder if chei's stat bonuses weren't too excessive too 01:44:25 48 01:44:49 as i brought up before 01:44:56 ponderousness is a +2 to delay 01:45:10 which scales down in amount of meaning as you add more 01:45:19 but the bonuses you get from it scale up as you add more 01:45:20 -!- syllogism has quit [] 01:47:12 +5 to all stats for going from delay 18 to delay 20 seems a bit extreme 01:49:05 also: hell hounds are silencable? 01:49:56 oh, and he grants resistances now too 01:50:00 i forgot about that part 01:53:28 -!- Spads has joined ##crawl-dev 02:05:39 no 02:05:42 I didn't mean to silence them 02:05:50 I was trying to Slouch 1 hit KO them 02:05:58 z b was silence, a b was slouch 02:10:36 -!- murphy_slaw_ has joined ##crawl-dev 02:23:30 hmm, I thought, somebody configured Henzell to post the dev blog's postings? 02:24:34 also: I don't want to say "told you so" but yesterday... I told you so 02:24:37 -!- Twinge- is now known as Twinge 02:25:06 I got away with so much bullshit that game because my stats were so huge 02:25:22 not to mention Slouch spam 02:25:53 also: I officially declare my retirement from using Chei.. 02:27:09 doy: didn't you add the dev blog to Ashenzari? (not henzell as reading enormous backlog showed) 02:31:42 // Slings are really easy to learn because they're not 02:31:42 // really all that good, and it's harder to get ammo anyway. 02:31:42 exercise(SK_SLINGS, 1 + random2avg(3, 2)); 02:37:53 Yeah, that could be looked into. :) 02:43:42 03by * rf101508ea25e 10/crawl-ref/source/directn.cc: Comment. 02:43:44 03by * rc803a3a7d84e 10/crawl-ref/source/message.cc: Reset msgwin.input_line when clearing messages. 02:49:22 -!- Vandal|PC has joined ##crawl-dev 02:51:27 -!- Vandal has quit [Read error: Connection reset by peer] 02:54:13 -!- Vandal|PC is now known as Vandal 02:59:39 Best pingback ever: http://www.smarttilework.com/?p=9834 03:01:40 for strange values of "best" :-D 03:03:27 That certainly is an odd pingback...but hey, who knows? 03:14:28 -!- Spads has quit [Ping timeout: 256 seconds] 03:16:24 -!- Spads has joined ##crawl-dev 03:23:53 what exactly is a pingback 03:29:15 sorear: blogging term, means that another blog has blogged about your blog posting 03:29:47 and that it automatically "pinged back" to let you know about it 03:33:58 e.g. a comment in the posting. Which I removed as spam. 03:39:43 bhaak, thought they pinged the server 03:39:45 the pingback shows up as comment, but it should be clearly marked as pingback so you know that it is a automatically generated notification 03:39:58 the blogosphere 03:39:59 ... 03:41:07 cbus: yeah, the term "pingback" is confusing, if you look at it with the knowledge what a network ping does 03:41:52 you have to clear your mind of knowledge to understand teh webkids 03:45:15 -!- by has joined ##crawl-dev 03:45:39 hey rob 03:45:51 morning due, ##crawl-dev 03:46:03 a lot on the web makes sense if you remember that computer people have NO knowledge of computer history and think they are always the first to use a generic term (e.g. "newsreader") -> lots of confusion and hilarity ensues 03:48:52 Woah. Do they actually say pingback instead of pong? 03:49:05 Hi by! 03:49:38 bah, I need a vimmeister. 03:50:17 due: ctrl-c ctrl-c ctrl-c :q! 03:50:24 ixtli: yepp 03:50:27 bhaak: No, the keyboard commant to switch between :split windows. 03:50:32 Is it ^w? 03:50:43 Ah, it is, ^w 03:51:40 I always forget and then use the : which I currently also forgot 03:51:48 due: you know the -p switch? 03:51:55 ^ww (to cycle) 03:52:01 ^w and then hjkl works 03:52:15 -!- scarf has joined ##crawl-dev 03:52:16 if you want to go in a specific direction 03:56:20 stock vim has a very unfortunate keybinding 03:56:34 ^w in insert mode = delete lots of text without saving it for the killring 03:56:55 :map! ^W ^O^W^W 03:57:06 if you have any intention of using windows in vim 04:07:19 hmm, doesn't happen here in gvim on windows and I don't know how to select stuff while in insert mode in the console version 04:07:21 should you be able to blink through trees? 04:07:36 (I tend towards "no", but unsure) 04:11:31 you should be able to blink anywhere in LOS 04:11:44 even through walls 04:11:47 if you ask me 04:11:59 the translucent wall rules are unneccessary and overcomplicated 04:12:09 and badly implemented, last time I checked 04:12:15 yes, it's horrible :) 04:12:31 they're also unintuitive 04:13:08 but I think it's not so bad if apportation/portal projectile/blink have consistent rules 04:16:37 apportation and portal projectile should work through glass too 04:23:18 sorear: I usually use tabs, actually, but I'm using split windows for the play-testing post. 04:32:13 Woo. Mixing Objective-C and C++ 04:33:20 bhaak, use v or V or ^V 04:34:01 did the monster energy thing get reverted? 04:34:52 monster pathfinding seems to be full of unnecessary optimizations 04:34:52 wasn't sticks to snakes level 1? 04:48:22 sorear: the translucent wall rules are rather intuitive to anyone who's used to D&D 04:48:29 as it's the same rules as there 04:48:35 but I don't see any reason why they /should/ be the same 04:48:43 Someone should tell dpeg that D&D has glass :D 04:48:56 greensnark: it has metal glass too 04:49:01 !tell dpeg Apparently D&D has glass too :P 04:49:01 greensnark: OK, I'll let dpeg know. 04:49:03 did sticks to snakes get changed from level 1 -> 2? 04:49:38 not since 0.3 04:49:39 greensnark: Epilogues don't support passing the level as a parameter to e.g. a milestone function ; due said that this'd require Summon Snark 04:49:50 ahh ok 04:49:55 Level as a parameter? 04:50:23 Heh, syllogism's Snake rune grab is awesome 04:50:24 : sewer_milestone(_G) 04:50:46 : sewer_milestone(_G) won't be in epilogue, no? 04:50:52 : always goes into the main map chunk 04:50:59 Putting that in epilogue{{ }} results in an empty "Lua error:" when the level is entered. 04:51:07 Huh, wot 04:51:16 Why is it empty :P 04:51:23 Let me try it here 04:51:26 greensnark: Yeah, but it should be, so it doesn't get marked multiple times if the map is rejected. 04:51:32 I have a patch, hold a second. 04:51:41 (with the milestones/epilogues, not a fix) 04:51:49 Oh, I agree that the milestone code should go into the epilogue 04:53:40 http://pastebin.ca/1805599 04:55:04 cbus: Sticks to snakes was non-present in 3.3 and present at level 2 in 4.0.0beta26 04:55:28 hmm, interesting, I've been actively trying to train transmutation as much as possible 04:55:30 greensnark: I think I fucked up in the error reporting code for the epilogue chunks. 04:55:37 and now by XP:7 my title was passed by dodging title 04:55:39 (on a DS) 04:55:51 guess its harder to train than fighting now :D 04:56:17 I'll get with the epilogue in a sec, watching syllogism on Shoals:5 :0 04:57:10 not worth using transmutation in trunk 04:57:17 too much hassle for too little pay off 04:57:37 Use slings instead, they're overpowered. 04:57:51 s/using/victory dancing/ 04:58:07 sorear, if you can't use the spells even with wizardry its not worth using 04:58:19 ooh 04:58:22 here's a thought 04:58:23 cbus: Don't have the heart to use Fulsome miscasts? I did. :) 04:58:28 keskitalo, me too 04:58:32 keskitalo, I got bored of holding down f10 04:58:59 greensnark: What if I make all melee attacks in forms train Transmigration? And make fulsome train Tmig a LOT because you can only use it once per corpse 04:59:47 Sounds worth trying 05:00:00 This syllogism rune fetch is ridiculous :P 05:00:00 yeah, might just be a problem with practising UC :) 05:00:06 It's like Charlie Chaplin 05:00:17 greensnark, what server? 05:00:22 FooTV 05:02:54 -!- scarf_ has joined ##crawl-dev 05:03:04 -!- scarf has quit [Read error: Connection reset by peer] 05:03:13 -!- scarf_ is now known as scarf 05:03:20 sorear, will need play testing but getting spider form castable might still be hard 05:03:24 (on low int chars) 05:03:39 !tell syllogism Nice Death's Door on Shoals rune :P 05:03:39 greensnark: OK, I'll let syllogism know. 05:04:36 Keskitalo: You shouldn't use : inside epilogue {{ }} 05:05:00 ..duh.. 05:05:02 cbus: not if you distill corpses 05:05:13 Desk, meet head. 05:05:23 sorear, been there done that :D 05:05:27 sorear, you run out of corpses pretty fast 05:05:38 cbus: I'm thinking of making distillation (all spells requiring materials) train 10-20 times faster than they do now 05:05:41 especially with low int when you actually need to eat a corpse every 2 casts :D 05:05:48 ahh ok 05:05:48 nice 05:05:53 in addition to having tmutted UC train it 05:06:06 yeah, sounds like a nice solution 05:06:14 no more need to victory dance like before 05:06:19 ok, first thing tomorrow... or second or third 05:06:24 * sorear -> sleep 05:06:27 and if you wanna practise UC you'll just stay out of shape 05:06:28 sleep well 05:09:40 greensnark: So my error handling is okay? 05:09:54 due: Looks like it's my bug actually :P 05:10:18 test_lua_boolchunk was being naughty and always reporting errors from the validate chunk :P 05:10:27 Instead of the actual chunk used 05:12:00 Bazaar welcome message doesn't work from inside an epilogue (although they don't need to be in one, right?) 05:12:17 Welcome message is not rigged to work from epilogue, no 05:12:25 But in epilogue you can just mpr it 05:12:36 Now we have epilogues we might as well remove welcome() 05:12:49 Yeah, sounds good. Could also add similar message to all the portal vaults. 05:13:07