00:53:42 valrus: You should just ask the git question, rather than asking about asking. :) 00:54:09 Enne, can I ask you a question? ;) 00:54:33 I got coolio to draw me a mouth tile and I'm planning on usering ti as an overlay on the relevant features to make monster tiles. 00:54:50 Sounds reasonable. 00:54:55 Which can then be used as the "known mimic" tile. 00:55:41 You can do this in rltiles with %start, %compose, and %finish. (If I'm remembering correctly.) 00:56:12 -!- Moriasc_ has joined ##crawl-dev 00:56:54 (Was that your question? Maybe I'm getting ahead of myself.) 00:58:31 Yeah, <3 overlays. 00:59:11 No, not really a qustion, that was a joke. 00:59:15 But then I rememberedI meant to tell you about the mouth thing. 00:59:18 -!- Moriasc has quit [Ping timeout: 272 seconds] 00:59:22 -!- Moriasc_ is now known as Moriasc 00:59:24 Ha. 00:59:33 The mimics sound neat. 00:59:43 At least you're adding tiles for these monsters. *tease* 01:00:00 :p 01:00:05 Did you see all the new monster enums I added? 01:00:10 I just haven't got around to implementing them yet. :> 01:06:39 -!- Moriasc has quit [Ping timeout: 252 seconds] 01:50:54 -!- Kurper has joined ##crawl-dev 02:33:19 -!- MarvinPA has quit [Ping timeout: 252 seconds] 02:52:21 -!- Kurper has quit [Remote host closed the connection] 03:09:32 -!- syllogism has joined ##crawl-dev 03:12:13 -!- hashc has quit [Quit: Ex-Chat] 03:13:58 -!- galehar has joined ##crawl-dev 03:14:01 morning 03:29:34 -!- monky has quit [Quit: hello] 04:25:22 hey, I noticed that with the new aptitude system, we're doing the same calculation over and over : 1 / exp(log(2) * apt / APT_DOUBLE) 04:25:25 is it expensive? 04:25:38 because it would be easy to save the results in a static table 04:25:57 there's only 11 aptitude values we're using 04:28:21 how often is it done? 04:29:01 I'd estimate about 50-300 ns for that expression on my hardware 04:31:27 It's done each time a skill is trained 04:31:55 It's trivial locally, I was wondering about server 04:33:00 well, don't 04:34:16 two trig function calls is utterly insignificant in 2010 04:34:44 especially on servers 04:34:53 ok 04:35:25 hardware trig support has been standard since the Pentium I 04:35:37 it won't take more than a couple ns 04:36:22 if yuo worry about performance issues, -> profiling 04:37:05 right 04:37:26 the profiles tell us that we're calling SHA256 hundreds of times per turn 04:37:47 you train skills less than once per turn on average 04:37:47 pow() may do the same a bit more efficiently 04:38:03 no 04:38:28 sorear: yeah, we should use some crypto quality RNG instead of hashing MT output 04:38:37 er 04:38:40 s/no/yo/g 04:38:55 * due does the mersenne twister with kilobyte 04:39:02 due: heh. And hi! 04:39:08 kilobyte: but if we did that, I wouldn't be able to wave sha256 in the face of everybody who frets about saving a few nanoseconds per turn! 04:39:44 What's the point of hashing MT output? 04:39:59 * due looks up MT to double check his understanding 04:40:27 prevents Adeon from cheating in the tourney 04:40:38 ah 04:40:51 due: you can calculate the whole internal state of MT from its output 04:41:07 And then predict future values. 04:41:13 kilobyte: yeah, I just noticed that 04:41:17 my 'ah' was directed at that. 04:41:27 MT is a fast and good RNG but it doesn't even try to be hard to reverse 04:41:36 if you see 624 values you can predict all future values 04:41:37 cool 04:41:53 why don't we just ban Adeon from the tourney? ;) 04:42:07 because anyone else can do the same 04:42:09 right. MT was created for simulations (which in a twisted sense Crawl is), not tournaments 04:42:36 -!- elliptic has quit [Quit: Leaving] 04:43:05 Banning Adeon was my original proposal, but we got a rather nice hardening patch donated 04:43:37 because nuking people who point out security holes is such a good idea... 04:45:06 Not nuking 04:45:29 banning 04:45:46 just... if you're going to go through the trouble to create a cracking system, and you brag about it, don't expect to have your scores taken seriously 04:45:47 I meant it in a wider context than just Crawl 04:46:08 since it's what countries and companies do... 04:46:30 how hard is to connect from a different IP with a different name? 04:46:38 Of course, the root problem was the decision to have real prises. 04:49:12 -!- Pseudonut has quit [Remote host closed the connection] 04:56:19 -!- dpeg has joined ##crawl-dev 04:56:21 cheers 04:56:43 hey 04:59:48 galehar: of course, I've been thinking last night about your question... 04:59:59 ... but now I have to read your new mail on the matter first :) 05:04:01 the basics of my new idea is: track skill points gained from crosstraining, and when reskilling, they are deduced from the skill, but not added to the other 05:04:19 well, the details are easier to code than explain... 05:04:26 that would have been exactly my proposal 05:05:03 when/how you rack up those bonus xp can be complicated (e.g. if the two skills are trained neck by neck) 05:05:21 but we shouldn't care about the order, but just track how much xp was gained from crosstraining 05:05:39 and for reskilling, start by removing this chunk of xp 05:05:56 it is not the most fair solution, but it's simple and will do the trick (avoid crosstrain abuse) 05:06:14 for real play, the matter is not so relevant anyway, so a solution like this will be good enough 05:06:38 galehar: Can we keep the aptitude boosts in this system? In the mail I was about to write, I reserved a paragraph for praising that idea :) 05:07:56 I believe it can be coded to be fair. crosstrained skill train faster, but are also drained faster when reskilling 05:08:15 okay, even better 05:08:37 As I said, the aptitude boost works, but the irony of it is that it is incompatible with reskilling :) 05:09:31 and anyway, it is very complex, so it would be hard to debug and maintain. And functionaly, it brings very little. 05:09:40 galehar: can you also show an aptitude malus for the casting skillsß 05:09:56 galehar: ah, so we don't use the aptitude changes anymore? 05:10:11 I had hoped that by tracking the xp saved from the aptitude boost, we can keep the idea. 05:10:13 I didn't commit it 05:10:39 oh, I know :) 05:11:11 galehar: but we can track the xp the player has been granted no matter whether it comes from the current mechanic or from the modified skill, no? 05:11:43 modified skill? 05:12:01 modified aptitude 05:13:04 but modified aptitude doesn't grant more xp. It lowers the amount of xp needed to reach a level 05:13:20 yes, but can't we keep track of that? 05:13:40 and really, I have spent days working on it and I'm sure of one thing: it's not a good idea 05:13:48 okay, I trust you 05:14:06 we have to keep the code as simple as possible 05:14:10 yes 05:16:36 why do the skills, practise_skill, skill_points and skill_order vectors all have a size of 50 instead of using NUM_SKILLS? 05:16:47 is it to make it easier to add new skills? 05:17:33 I don't know. 05:17:55 well, maybe this wasn't really addressed to you ;) 05:28:14 galehar: because hard-coded variables ftw. 05:28:24 possibly hangers-on from when Magic Numbrers were Sexy. 05:29:41 ok, I see what you mean 05:29:56 but I'm not sure how to fix this without busting save compat 05:30:06 it shouldn't 05:30:12 actually, in theory, it can't 05:32:16 but NUM_SKILLS != 50 05:33:09 what is it? 05:33:18 number of skills 05:33:23 okay 05:33:27 which is like 30 or whatever 05:33:33 yeah, something like that 05:33:34 the point is, check all th eiterations of skills/practice_skill/etc 05:33:42 they can't exceed NUM_SKILLS 05:33:53 the problem is with save file 05:34:12 loading them? 05:34:15 yes 05:34:18 ahh 05:34:29 just inc minor version and put a check in and dump anything greater than num_skills 05:35:26 I have to inc minor version anyway because I'm adding a new vector, so I might as well fix that will I am at it 05:51:36 hmm... did you already do this? 05:52:25 I wanted to mess with this exact problem during the ench split, but preferred not to in order to not introduce something that might be hard to merge 05:53:05 but then, there were few to none constructive comments about the ench split, so I may just as well push it to trunk 05:53:37 well, I am doing it 05:53:50 but I can cancel 05:54:04 as I said, I need to add a vector anyway 05:55:01 due, dpeg: is it ok to push the current version of the ench split? 05:57:24 or actually, merging the tag reading code can't be that complex 05:57:34 how did you split it? commit message was... concise 05:58:06 even if it's about inserting a skill in the middle (as magic skills need to be together, not separated by Invo and Evo) 05:58:20 galehar: the current version is Hexes vs Charms 05:59:03 this is the proposal that made the most sense to me, and the discussion afterwards didn't bring anything new 05:59:28 ok 06:02:49 so go ahead, I'll merge the tag reading skill so there's no need to hurry 06:03:08 ok 06:19:10 galehar: about the mail: appears to be ok 06:21:51 -!- MarvinPA has joined ##crawl-dev 06:33:20 -!- MarvinPA has quit [] 06:37:41 -!- MarvinPA has joined ##crawl-dev 07:00:55 how is the new request-less mantis doing? is it better than before? 07:01:18 I saw a lots of issues in the "Patches" category, which seems to have taken off nicely :) 07:16:21 yues, it is nice 07:16:39 Napkin: if we ever switch trackers, i'm recommending redmine 07:16:41 it's fucking *amazing*. 07:16:57 getting the gitosis plugins to work is a fucking hellish nightmare that i still haven't gotten properly sorted 07:17:18 i saw redmine, it was quite nice 07:17:26 I've got it running on my work intranet 07:17:30 but it's also a terribly huge construct 07:17:43 and.. in ruby on rails, right? 07:18:00 and providing much more than we actually needed for crawl ;) 07:18:51 it does bug tracking, git integration and wiki 07:18:52 i still haven't looked at gitosis, to be honest 07:19:01 i still don't understand what gitosis *is* 07:19:15 oh, gitosis is a user-management script 07:19:19 oh 07:19:29 then i fucked up the installation :) 07:19:42 definining read/write permissions based on public ssh keys 07:19:50 oh interesting 07:19:57 while still using only one real unix account 07:20:10 isn't that easily done with authorized_keys, though? 07:20:18 I suppose not 07:20:32 with authorized_keys one can easily map users to git users... 07:20:36 or at least, i think i had something lik ethat seet up 07:21:20 maybe that was svn 07:22:35 that's what gitosis does, yes ;) 07:22:43 but only to one unix account 07:22:43 oh. right. 07:23:02 so you don't have to create any additional accounts, just install the key 07:45:09 03dolorous * r1e37b1cd882c 10/crawl-ref/source/dat/des/portals/wizlab.des: Add minor cosmetic fixes. 07:45:43 -!- eith has joined ##crawl-dev 07:49:04 hmm, could I really be inside a vulcano which has no loot at all? 07:49:10 03dolorous * rde13dc7f5439 10/crawl-ref/source/mapdef.cc: Typo fix. 07:51:37 due: withdrawn turtles are a pain to kill with low unarmed and no slaying, but freezing cloud kills them nearly instantly 07:51:51 as opposed to, say, lightning bolt which needs 1e38 casts 07:52:52 a similar issue as with new Agnes, except there it's EV not AC. Even lev1 freeze > lev6 bolts of cold, etc. 07:54:16 -!- upsy has joined ##crawl-dev 07:56:56 -!- Twinge has quit [Ping timeout: 255 seconds] 07:58:28 back 07:58:31 kilobyte: yes, it's okay 07:58:48 kilobyte: as usual, there'll be more comments (also lourder :) after commit 08:06:23 hm 08:06:55 Napkin: i was referring to having one unix account with a variet of commands like 'git --user =blah' per key 08:07:03 Napkin: but i remember now that this was my svn setup 08:08:13 -!- Noom has joined ##crawl-dev 08:09:01 -!- blackpenguin has quit [Quit: .] 08:11:06 good thing you reminded me, due 08:11:12 i'll have a look now :) 08:11:21 otherwise i'll fall asleep today.. 08:16:22 03dolorous * r56ff0e94d1aa 10/crawl-ref/source/mon-act.cc: Fix indentation. 08:17:49 sleepbad ;) 08:27:11 -!- Cryp71c has joined ##crawl-dev 08:31:24 hmm, there is something new and shiny called gitolite 08:31:29 will read up on that 08:37:22 -!- Mu_ has joined ##crawl-dev 08:50:46 -!- Moriasc has joined ##crawl-dev 09:58:58 -!- elliptic has joined ##crawl-dev 10:04:05 03dolorous * r4bb882590064 10/crawl-ref/source/dat/des/portals/ (sewer.des volcano.des): Remove now-redundant setting of both name_adjective and name_descriptor. 10:04:09 03dolorous * ree55a365e89b 10/crawl-ref/source/mon-death.cc: Add minor cosmetic fix. 10:04:12 03dolorous * rf3d007255525 10/crawl-ref/source/ (mapdef.cc mon-info.cc): Reorder monster name override logic a bit in mapdef.cc and mon-info.cc. 10:04:13 03dolorous * r45292ff96e41 10/crawl-ref/source/ (mapdef.cc mon-info.cc): Attempt to fix monster name override regressions (various Mantises). 10:40:37 -!- Zaba has quit [Ping timeout: 252 seconds] 10:48:56 03dolorous * r5b5b0abf8048 10/crawl-ref/source/itemname.cc: Typo fix. 10:49:01 03dolorous * rbb2aa63d001b 10/crawl-ref/source/mon-util.cc: Tweak comment. 10:49:03 03dolorous * r92fda6dbb253 10/crawl-ref/source/mon-util.cc: Fix Mantis 2596. 10:51:20 -!- Zaba has joined ##crawl-dev 10:55:05 -!- Moriasc has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630]] 11:07:25 -!- blackpenguin has joined ##crawl-dev 11:37:49 Draconian drops skills at XL7 in a confusing manner (https://crawl.develz.org/mantis/view.php?id=2812) by S2im 11:47:53 -!- monky has joined ##crawl-dev 11:52:36 -!- enne has quit [Read error: Connection reset by peer] 11:52:55 -!- enne has joined ##crawl-dev 12:02:22 -!- galehar has quit [Quit: Page closed] 12:07:46 -!- Kurper has joined ##crawl-dev 12:18:03 Control F level preview re-prompts "entering level" alerts. (https://crawl.develz.org/mantis/view.php?id=2813) by Twilight 12:22:07 -!- Twinge has joined ##crawl-dev 12:23:05 Control F level preview re-prompts "entering level" alerts. (https://crawl.develz.org/mantis/view.php?id=2814) by Twilight 12:32:59 -!- elliptic has quit [Quit: Leaving] 13:11:22 -!- Pseudonut has joined ##crawl-dev 13:22:50 -!- Pseudonut has quit [Remote host closed the connection] 13:59:56 -!- TGWi has joined ##crawl-dev 14:05:19 dpeg: there's no problem with ew i-iii and ea uncursing items 14:05:28 I actually used them intentionally to uncurse things because scrolls of rc are in short supply 14:06:03 curse weapon and curse armour are common, and you don't even need all your armour cursed - you can leave your robe uncursed until it's fully enchanted, for example 14:06:09 and still be bound 14:07:28 TGWi: rc => curse should be paper => curse, c/d 14:10:09 -!- Pseudonut has joined ##crawl-dev 14:31:28 m1nced (L14 KeFE) ASSERT(probe != attack_constraints.connection_constraints->end()) in 'mon-abil.cc' at line 1569 failed. (Shoals:2) 14:32:28 m1nced (L14 KeFE) ASSERT(probe != attack_constraints.connection_constraints->end()) in 'mon-abil.cc' at line 1569 failed. (Shoals:2) 14:32:30 m1nced (L14 KeFE) ASSERT(probe != attack_constraints.connection_constraints->end()) in 'mon-abil.cc' at line 1569 failed. (Shoals:2) 14:40:26 -!- MarvinPA has quit [] 14:54:26 -!- TGWi has quit [Read error: Connection reset by peer] 14:55:20 -!- TGWi has joined ##crawl-dev 15:09:58 -!- Siber has quit [Quit: Whoa! Partial hit!] 15:13:54 -!- valrus has joined ##crawl-dev 15:21:32 is there an awesome and cool and rad person in here who can help me wrestle with git to produce a patch from my code? 15:55:30 back 15:55:38 valrus: I don't qualify, sorry. 15:56:21 oh well 15:56:34 -!- Cryp71c has quit [Quit: Leaving] 16:05:39 -!- galehar has joined ##crawl-dev 16:05:57 evening 16:06:05 Hi 16:06:12 21:21 < valrus> is there an awesome and cool and rad person in here who can help me wrestle with git to produce a patch from my code? 16:06:17 galehar: ^ :) 16:06:26 sure 16:06:32 valrus: still around? 16:06:35 yeah 16:06:40 thanks dpeg 16:06:48 you commit your change 16:06:54 done that 16:07:00 and run git format-patch -1 16:07:07 I have 2 or 3 possible problems 16:07:27 I had a couple intermediate commits 16:07:42 do you want to merge them all in one patch? 16:07:43 does that matter? 16:07:53 yes 16:07:57 also 16:07:57 well, you can either git format-patch -3 16:08:02 which makes 3 patches 16:08:11 or git reset HEAD^ 16:08:16 several times 16:08:20 and recommit 16:09:28 ok. also... I have an old commit from my last patch which has since been incorporated into origin/master (I think) but lingers in my local working branch 16:10:33 well, if it has been commited without any change, it shouldn't be a problem 16:10:33 and the basis for my patch is probably a couple weeks old, I don't know if that's a problem 16:10:42 well it can be 16:10:50 better start with something fresh 16:11:10 how can I do that? my work's been in progress for a couple weeks 16:11:21 if you have set up git properly with all the options described in the doc, you can just git pull 16:11:22 merge it occasionally along the way? 16:11:41 you'll get all the updates, and then your local commits will be applied on top of the new head 16:12:24 how do I make that happen 16:12:30 just git merge? 16:13:03 you have your patches in a branch? 16:13:12 yes 16:13:28 ok, so git checkout master 16:13:32 git pull 16:13:38 git checkout your_branch 16:13:39 done those today 16:13:42 git rebase master 16:14:05 will do what you want 16:15:54 -!- Wensley has joined ##crawl-dev 16:16:40 ok, I appear to have (at some point while trying to do what I want on my own) killed all my changes to at least main.cc 16:16:57 :( 16:17:01 03galehar * r30c0d2fc1081 10/crawl-ref/source/invent.cc: Fix wrong items showing up in the wield screen 16:17:02 03galehar * rf85fe49358e8 10/crawl-ref/source/ (4 files in 2 dirs): Add Omndra new title screen 16:17:06 03galehar * r27f95f222745 10/crawl-ref/docs/develop/save_compatibility.txt: Small update in the save_compatibility doc 16:17:06 well I still have all the patch files 16:17:07 03galehar * r9d8bd352a1d3 10/crawl-ref/source/tilereg-title.cc: Fix crash with gdb and the new random title screen code 16:17:13 so I think it's not the end of the world 16:17:40 well, best thing to do is create a new branch from the fresh master and apply your patches there 16:17:54 check if everything's ok 16:17:55 ok 16:18:07 how do I apply a patch 16:18:13 git apply patch_file 16:18:22 what are you coding btw? 16:18:28 new time step 16:18:36 ok 16:18:58 when you apply the patches, it doesn't commit 16:19:00 valrus: were your changes committed? If so, they're kept for a long time (1-3 months depending on their type), "git reflog" will tell you the hash. 16:19:31 so you can apply all of them, and commit once and then create a single patch file 16:20:08 wait, I'm stupid, my changes are still here, I was just searching for the wrong term to look for them 16:20:14 :) 16:20:19 apparently none of my changes involved the string "CHEI" :P 16:20:50 use -i in [git] grep :p 16:21:21 it was in vim, and I think it was case-insensitive 16:22:42 urgh I hate being a noob. I think I was in the middle of a rebase in my branch when I pulled master and now I can't abort my rebase :/ 16:23:40 applying the patches to a new branch sounds like a good option at this point 16:25:40 I get a lot of "patch does not apply" 16:31:10 by default, it is very susceptible. Use --reject 16:31:43 it will merge what it can and put rejected hunks in *.rej 16:32:16 I reset master so I could cancel the rebase 16:36:16 galehar: thanks for pulling the new starting screen! 16:37:00 adding new ones is now as simple as putting the file in the directory 16:37:21 yes, I saw, that's cool too 16:38:41 galehar: something that I learned along the way is that often skewed (i.e. non flat) probability distributions are better when it comes to random content. E.g. when you have five maps for the same portal vault, make two common, two rare, one very rare. Sane with speech. Can we have something similar with starting screens? (This will only really matter once we have many screens, of course :) 16:39:13 sure, we can add that later 16:39:28 I was thinking of putting the weight right in the filename 16:39:39 like title_w_xx.png 16:39:52 or something like that 16:39:58 cool... for later 16:41:17 ok. galehar: I've successfully rebased my branch. do I just need to format the patch now? 16:41:42 check git log 16:41:46 and yes 16:42:04 I think there's still 3 patches because of my intermediate commits? 16:42:31 03kilobyte * rda93ffd8693b 10/crawl-ref/source/misc.cc: Don't consider places next to slime walls as "safe". 16:42:38 neither of which I think is useful for anyone 16:43:06 any Java fans in here? 16:43:10 do git reset HEAD^ 16:43:21 what exactly does git reset HEAD^ do? it sounds dangerous 16:43:29 actually I guess it's just the ^ part 16:43:35 that I don't understand 16:43:54 reset --hard is very dangerous 16:43:59 will cancel the latest commit but without actually touching the file 16:44:35 is HEAD^ the last commit before HEAD? 16:44:45 exaclty. reset --hard will revert the change. 16:44:49 yes 16:44:56 ok. 16:44:58 and HEAD^^ the one before 16:45:10 and HEAD~3, well you guess :) 16:45:33 :) 16:45:56 ok, down to one commit. should I reset to before that one too and then just do a new commit? 16:46:01 or rename the remaining one? 16:46:26 reset the last one and recommit 16:46:30 ok 16:47:02 there is surely a more elegant way of doing this, I'm just learning git myself 16:47:29 it's fine, I'm an utter beginner and this is extremely helpful. almost surely more helpful than if you'd been an expert 16:48:07 well, it's good to start with the basic functions :) 16:48:18 yeah 16:48:32 most git references online seem to assume commit access 16:49:05 to origin, or maybe I was just incorrectly reading that into it! 16:49:22 well, have you read docs/develop/patch_guide.txt? 16:49:43 the latter seems more likely now. until recently the idea of a "local commit" was odd to me, which shows you how severe my misconceptions were ;) 16:49:51 and more importantly, have you read docs/develop/coding_conventions.txt? 16:50:12 I believe I've read both, though I'm not sure I could pass a pop quiz right now 16:50:24 my first patch yielded no code complaints 16:50:29 good :) 16:52:41 I have just tested it, to merge 3 patches, instead of doing git reset HEAD^ 3 times, you can just git reset HEAD~3 16:52:53 that makes sense 16:57:18 yay I have a patch! 16:57:25 thanks galehar, you are the best ever 16:57:47 * dpeg congrats valrus & galehar. 16:57:52 that was try 2, maybe next time I can do it on my own :) 16:58:22 I guess I should make sure it still compiles before I submit it :P 17:01:20 -!- Moriasc has joined ##crawl-dev 17:10:30 no Java fans? 17:10:46 Java has fans? 17:11:05 I prefer cocoa over java. 17:11:22 I think Java would be an improvement on what we have now. 17:11:39 I'm learning Java for a college class 17:11:55 sorear: you really hate c++, don't you? 17:12:03 I'm a functional programming fanatic 17:13:27 no php fans? 17:13:37 php has fans? 17:13:40 I like php ok? 17:13:54 galehar: c++ is a great language for embedded systems work. in the mainstream arena nothing else comes close to C++'s handling of real time and real memory constraints 17:14:34 stuff like POWDER (32 kb ram and an 8MHz arm core on the GBA) couldn't exist without it 17:14:43 but aren't all the commercial games coded in c++ too? 17:15:23 well, porting crawl to handheld devices is one of our goals 17:20:12 * jld is hating C++ lately, but a lot of that is due to how it handles concurrency 17:20:18 (i.e., it doesn't) 17:21:35 languages are just tools 17:21:51 you just need to chose the tool appropriate to your work 17:22:44 haha galehar I only just saw your comment on time step on the wiki. guess I won't submit my patch just eet 17:22:46 *yet 17:23:17 I don't even remember it :) 17:23:20 let me check 17:23:27 https://crawl.develz.org/wiki/doku.php?id=dcss:brainstorm:god:cheibriados 17:24:25 you mean about draining piety? 17:24:29 yeah 17:24:59 well, dpeg said to me that the mechanism already exist with yred pain mirror 17:25:02 can you post the comment here? 17:25:12 "How about a fixed cost, and an invocation dependent duration? It is more consistent with all the other god abilities. The problem with draining piety is that the piety counter is not precise. So you have no idea how long it is going to last or at what rate your piety is decreasing." 17:25:39 galehar: yes, the remark is spot on, but there's still a precedent with Injure Mirror. 17:25:53 the yred mechanism is different because piety drain is based on damage taken 17:26:08 I'm not sure how differently the player will perceive it 17:26:36 Yrdy's is more difficult to grasp than Che's 17:27:04 you mean how fast your piety goes down? 17:27:18 yes 17:27:37 is there a warning when the ability is about to expire? 17:27:40 sorear: C++ on a 32K system? It's a heavily wasteful language -- you can write memory-efficient code in C++, but that's by using a subset which is basically C. 17:27:45 currently, no 17:27:49 should be though 17:27:58 of course, it's orders of magnitude better than php or java here, of course 17:27:58 no idea why I didn't think of that 17:28:05 kilobyte: hi, new Ash todo on the wiki page :) 17:28:19 yes, either with draining piety or invo dependent duration, you have to warn the player when it's about to expire 17:28:21 kilobyte: the bits about reskilling will be taken care of by galehar, afaiu 17:28:21 also, currently invo only affects success rate, which seems wrong 17:28:38 valrus: Ash has no use for Inv 17:29:05 dpeg: yes, I'm still working on the crosstraining abuse and will do the rest after 17:29:09 well, neither does, say, veh 17:29:28 but are there any gods whose highest active ability doesn't depend on invo? 17:29:37 trog, nemelex 17:29:39 not counting trog, nemelex 17:29:43 wiseass 17:29:45 dpeg: being able to reskill to 0 skills was actually a request from kilobyte :) 17:30:08 monky I was totally typing that already when your msg appeared :P 17:30:09 kilobyte: I don't think it's a good idea. If we drop that, we can give players the m screen they're familiar with. 17:30:15 you want to reuse the m screen for interface consistency? 17:30:26 And it's not like it is hard to get some skill to 1. 17:30:29 galehar: yes 17:30:53 valrus: There is no way to train Inv with Ash, so no ability should check Inv. 17:30:55 ok, I'll look into it 17:31:22 right. but Chei's abilities all use inv, so it should matter for the final ability 17:31:28 (it seems to me) 17:31:28 yes 17:31:46 valrus: ah, you're talking Che, not Ash. For the slow god, Inv is okay to use, of course. 17:31:51 yeah 17:32:13 I'm probably going to change new time step to galehar's idea 17:32:16 dpeg: my reason was that some skills auto-fail at 0, especially most uses of Evo 17:32:39 but that can be fixed by changing the success formula there 17:32:43 kilobyte: it's easy to train evo with wands 17:32:50 kilobyte: ah. But that's a deficit with the skills and should be repaired there... 17:33:09 flat piety cost (with add'l penalties for using it too rapidly) and invo-dependent duration 17:33:18 rather than draining piety while in use 17:33:40 valrus: that's okay, too. 17:33:49 kilobyte: please explain how the language itself is wasteful 17:33:56 * galehar 's graphic card broke down and he's compiling on his wife's netbook, so he has a lot of time to chat on IRC tonight :( 17:34:00 The higher piety cost for repeated use might be hard to convey... it's really a mechanic not used before :) 17:34:12 kilobyte: the STL sucks, I'll grant that 17:34:36 galehar: no ssh daemon on the headless box? 17:34:40 galehar: be happy your wife has one 17:34:49 dpeg: it might not be necessary, with the piety cost up front 17:35:07 sorear: yeah, templates were my #1 argument. C++ does a lot of mallocs behind your back, it's hard to control it when you use anything object related. 17:35:20 sorear: it won't start without a graphic card (there's no integrated one) 17:36:01 but currently I have a message like "Cheibriados is [adverb] annoyed with your hastily jumping in and out of time!" when the additional piety penalty is incurred 17:36:09 it's bad enough that for example C++ bindings for GTK take more space than C# ones 17:36:27 galehar: gee that's right up there with "No keyboard, press F1 to continue" 17:36:36 :) 17:36:44 or, compare the memory usage of Crawl vs NetHack 17:37:21 valrus: sounds okay. Please inform me (best via email) about whatever you do. 17:37:29 kilobyte: what about it? Is Crawl so inefficient? 17:37:33 kilobyte: please give examples of uncontrollable mallocs 17:37:47 dpeg: ok 17:37:51 * galehar wonder if we should program the bots to kick anyone who says the nh word 17:38:03 galehar: I think we can persuade kilobyte to relax on reskilling 0 skills. 17:38:09 the keyboard warning makes sense... those who will use the box headless have enough clue so for them it's just a mere annoyance, while typical users would be completely stumped 17:38:11 galehar: Why? 17:38:11 if you use operator new, you get a heap object 17:38:35 dpeg: would be funny! 17:38:51 sorear: not "uncontrollable", rather "insidious if you're not especially careful" 17:39:21 galehar: for better or worse, NH is the champion in the field. 17:40:27 dpeg: I'm talking memory and CPU usage; for CPU Crawl is excused as it does a lot of complex stuff NetHack does, but memory wise, not so much 17:40:30 The bots should be programmed to kick anyone who says something good about nh. 17:41:01 kilobyte: but our levels are much bigger, with many more monsters etc.? 17:41:23 more importantly, the travel cache 17:41:39 nethack doesn't keep information about every item in the dungeon in memory at once 17:42:28 sorear: certain simple objects are placed on the stack, but a good majority is malloced. About anything that's not a simple object of a known type and no variable-size parts. 17:42:31 kilobyte: What do you think the best language for a Crawl-like program is? 17:43:09 Immediate follow-up: Could Linley have known when starting with Dungeon Crawl 0.1? 17:43:54 sorear: not sure, but the arguments I mentioned before don't really apply to _this_ question. Memory efficiency doesn't matter much on servers with multiple GB RAM. 17:44:21 it's not that 32K box you mentioned 17:44:23 kilobyte: 1 GB / 50 users during tournament peak 17:44:33 Crawl has memory issues 17:44:45 nowhere near as bad as POWDER's; 17:45:27 I am asking you, independant of the previous discussion, what you think the best tool would be. 17:45:30 CDO uses a hard ulimit of 100MB and it's hit only for infinite loops now that the memory leaks for database build are fixed 17:46:06 -!- Moriasc_ has joined ##crawl-dev 17:47:10 sorear: it depends on the dev team. For a single coder project, I'd pick C, but for a large team, C++ is more manageable. 17:47:38 What's POWDER written in? C or C++? 17:47:40 C++ 17:48:41 -!- Moriasc has quit [Ping timeout: 255 seconds] 17:48:47 -!- Moriasc_ is now known as Moriasc 17:52:03 sorear: it's an efficiency vs ease of writing issue. Ie, exactly what is described by the low-vs-high level language concept. 17:54:11 -!- Moriasc_ has joined ##crawl-dev 17:55:03 with today's game complexity and hardware, it's no wonder that almost all games are coded in C++ 17:55:35 kilobyte: You don't think there are any languges that would help with "ease of writing"? The 10 minute compile cycles get to me 17:55:57 * dpeg has recently discoverd the power of -j4 17:56:20 * galehar has ordered a quad core with his new graphic card 17:57:14 -!- Moriasc has quit [Ping timeout: 255 seconds] 17:57:21 -!- Moriasc_ is now known as Moriasc 17:57:50 sorear: indeed C++ has an issue with compilation stuff, but then, any modern machine can do a debug build in 1m 40s or so 17:57:56 s/stuff/speed/ 17:58:44 -!- pointless_ has joined ##crawl-dev 17:59:07 pointless_: hi 17:59:10 hi 18:00:54 Crawl's rebuild are of two types: those with any common .h file touched, and those without. The former cause a near-total rebuild and ccache is of no help. 18:01:18 edit enum.h and due will scream at you since he codes on a netbook 18:02:00 pointless_: meow! 18:02:10 dpeg: does 35 turns sound too long for a duration of time step at 27 invo/200 piety? 18:03:04 kilobyte: hi, sent you an e-mail 18:03:09 my machine is only slightly faster than due's netbook and I gave up on following after people kept editing enum.h 18:04:02 well, we have to mess with include files from time to time 18:04:29 valrus: yes, I'd say so 18:04:46 how about we put a donation button on our website to buy cpu for the devs :) 18:04:58 pointless_: thanks! 18:05:00 what's a decent max? 20-ish? 18:05:04 what happened to ] ? 18:05:05 galehar: Jeff told me we'd make more money than I expect :) 18:05:13 pointless_: the key command? 18:05:17 yes 18:05:29 pointless_: well, only few use it. What happened to it? 18:05:51 I don't know, I'm pressing it and it says "You aren't carrying any such objects" 18:05:55 valrus: 5-15, I'd say 18:06:05 I think it used to bring up currently equipped armor + jewellery 18:06:08 ok 18:06:09 pointless_: what happens if you press [ ? 18:06:28 displays current armour slots, so that is unchanged 18:06:44 valrus: 35 allows you to dash for the orb after the first chamber 18:06:58 perhaps that's okay, I don't know 18:09:10 sorear: the machine I sit on is an ancient P4-era Celery that takes 30mins for a debug build and I don't even bother upgrading it since all I do on it is browsing and sshing to actual devel boxes... 18:09:41 and there are zounds of servers with totally unused CPU around 18:09:55 ok now ] is prompting me to select a hand weapon since I picked up a dagger 18:10:33 people seem to don't go below four core for a file server that could just as well be a 486 with no speed loss 18:11:31 well, Dell won't sell them any less, that's why :) 18:15:57 oh help says ] shows currently equipped items, I guess I never tried using it with nothing equipped before 18:16:18 but I don't know why the dagger was auto-quivered when I can't throw anything 18:18:50 pointless_: butchering? 18:19:03 dpeg: ok, I have a formula that gives an effective range of about 5 to 16 18:19:11 it got put in my quiver slot, although I cant throw it 18:19:44 the low end might be a little higher depending on at what invo level it actually becomes usable 18:20:13 ok 18:20:43 -!- Wensley has quit [Ping timeout: 240 seconds] 18:31:26 You feel extremely strange. The horns on your head shrink away. A pair of antennae grows on your head! Your thoughts seem clearer. 18:31:31 yay on an Ashite... 18:32:24 kilobyte: yes, I had that too 18:33:14 the Elf coloring may be only slightly goofy in most cases, but when you get Scrying on, it's just ridiculous 18:35:14 I guess we'll do something about it before release. Something can surely be done. 19:01:20 -!- dpeg has quit [Quit: bye] 19:04:27 wizmode: raised fighting, repeatedly reduced my hp and borged myself, ash skill transfer, -5/-5 hp 19:05:19 :) 19:05:26 killed by greed 19:05:42 not dead but I die from taking any damage 19:05:58 even better 19:06:15 so you reskilled from fighting? 19:06:20 galehar: what' 19:06:27 yes 19:06:27 s your opinion on eronarn's proposal? 19:06:57 for what it's worth, you can also get negative maxhp by doing DDoor+draining after borging a few times 19:07:22 monky: will look into it 19:07:35 tgwi: don't really have any 19:07:43 might be good, might be not 19:08:00 the guy's attitude doesn't really makes me want to look into it 19:08:09 and certainly not coding it 19:12:44 -!- Pseudonut has quit [Remote host closed the connection] 19:12:58 -!- syllogism has quit [] 19:15:32 -!- Moriasc_ has joined ##crawl-dev 19:15:34 The deep elf knight shoots an arrow. The arrow misses you. The arrow hits Boris. Boris is destroyed! 19:15:46 yay procrastinating for that xp stealing thingy 19:17:20 -!- Moriasc has quit [Ping timeout: 255 seconds] 19:17:27 -!- Moriasc_ is now known as Moriasc 19:24:08 03kilobyte * r5a47fe6fe7d9 10/crawl-ref/source/misc.cc: Felids: remove silence when Boris smotes your sorry ass. 19:35:35 -!- TGWi has left ##crawl-dev 19:36:27 is the haste nerf on cdo? 19:40:43 is there any way to check how long god wrath lasts in my game? 19:42:57 pointless_: yes 19:43:39 Noom: ^O will give you a two-point information -- the god's name is either light or dark red 19:43:46 03galehar * ree4b7d2665ca 10/crawl-ref/source/ (9 files): Fix the crosstraining abuse with Ash's reskilling 19:43:58 Noom: for a more accurate info, you may look at an altar 19:44:14 nm just checked 19:44:20 using !lm 19:47:30 -!- galehar has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630]] 19:49:59 03kilobyte * r0f5ed2bd6d36 10/crawl-ref/source/misc.cc: Force a more on death. 19:50:01 03kilobyte * rdda5d09bfbde 10/crawl-ref/source/monster.cc: Haste nerf: adjust the speed of Pan lords, but by less than others. 19:54:49 -!- OG17 has quit [Read error: Connection reset by peer] 19:55:29 -!- eith has quit [Ping timeout: 276 seconds] 19:57:08 -!- Pseudonut has joined ##crawl-dev 20:05:24 -!- OG17 has joined ##crawl-dev 20:15:03 03kilobyte 07ench-split * rb69664f23318 10/crawl-ref/ (82 files in 9 dirs): Merge branch 'master' into ench-split 20:30:53 Crash on doll selection screen (https://crawl.develz.org/mantis/view.php?id=2815) by moxian 20:31:37 -!- TGWi has joined ##crawl-dev 20:31:50 -!- Mu_ has quit [Quit: Defecator, may everything turn out okay so that you can leave this place.] 21:05:13 -!- pointless_ has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630]] 21:07:08 -!- Kurper has quit [Remote host closed the connection] 21:44:34 -!- upsy has quit [Quit: Leaving] 21:51:07 -!- Noom has quit [Ping timeout: 240 seconds] 22:32:23 -!- valrus has quit [Remote host closed the connection] 23:11:55 -!- paxed has quit [Ping timeout: 240 seconds] 23:29:27 -!- paxed has joined ##crawl-dev