--- Log opened Wed May 23 00:00:33 2018 20180523 00:01:12<+discordbot> jyrki says no app is supposed to use text input for hotkeys, so I believe him that we're doing something wrong. 20180523 00:01:29<+discordbot> text input events* 20180523 00:01:57< gfgtdf> noone was yet able give me a good reason why we use sdl text input, back when i asked aginor he/she just said that other people insisted on it. 20180523 00:02:25<+discordbot> I don't understand it either 20180523 00:03:15<+discordbot> @Vultraz I meant to map =. On laptops (I get not everyone has this keyboard mapping…) many folks have + as shift = . 20180523 00:03:19<+discordbot> Aginor complained a lot about that back int he day. 20180523 00:03:24< celticminstrel> IIRC I blacklisted the numeric keypad from the textinput hotkey system. 20180523 00:03:43< celticminstrel> So that they can be bound separately from the digits and such. 20180523 00:03:47<+discordbot> He actually changed the system we use for hotkeys in 1.13.x but at some point that appears to have been reverted to be more like 1.12.x. 20180523 00:04:02<+discordbot> So we have a system that everyone agrees is the wrong way to go and yet none of us know why we have it 20180523 00:04:02< gfgtdf> what exactly is 'that' ? 20180523 00:04:04<+discordbot> lovely 20180523 00:04:28< gfgtdf> 1.12 didn't use textinput though. 20180523 00:04:40<+discordbot> celmin: do you know why we use textinput 20180523 00:04:42<+discordbot> The thing is the system that is purportedly "right" didn't feel right to me as a user. 20180523 00:04:53< celticminstrel> I think the reason why the hotkey system uses textinput is so that (for example) : can be the hotkey for the command console even if that doesn't happen to be shift+; 20180523 00:04:55<+discordbot> Because suddenly I was required to type ñ to access the command input. 20180523 00:05:00<+discordbot> Instead of :. 20180523 00:05:11<+discordbot> (Which is Shift+. on my keyboard.) 20180523 00:05:30< irker017> wesnoth: gfgtdf wesnoth:1.14 d1097d93d5a4 / src/playmp_controller.cpp: fix oos when using 'back to turn' in mp https://github.com/wesnoth/wesnoth/commit/d1097d93d5a41eee7cd373ad43a0a9af32c48fd1 20180523 00:05:49< celticminstrel> The "right" way apparently is to use scancodes, but I'm not sure you could get this effect by doing so. 20180523 00:06:07<+discordbot> Yeah, that's what I am moaning about. 20180523 00:06:16< celticminstrel> Oh yeah, did 1.14.2 get tagged yet? 20180523 00:06:17<+discordbot> The right way felt completely confusing to me. 20180523 00:06:19<+discordbot> celmin: so that's the latter behavior I described. The same key, regardless of the combination that produces it 20180523 00:06:20<+discordbot> No. 20180523 00:06:45< celticminstrel> @Vultraz What? 20180523 00:06:51<+discordbot> ........ 20180523 00:06:54< celticminstrel> What you said doesn't make sense. 20180523 00:06:57<+discordbot> I asked 20180523 00:07:00<+discordbot> just afew minutes ago 20180523 00:07:11< celticminstrel> "The same key, regardless of the combination that produces it" does not make sense. 20180523 00:07:12<+discordbot> "Vultraz - Today at 10:56 AM are hotkey supposed to be a combination, regardless of what it produces on any keyboard, or are they supposed to be specific keys, like "+", regardless of the combinations that produce them on any keyboard" 20180523 00:07:34< celticminstrel> + is not a specific key. 20180523 00:07:41<+discordbot> I use "key" loosely 20180523 00:07:44< celticminstrel> (Unless you're referring specifically to the key on the numeric keypad.) 20180523 00:07:48< celticminstrel> Yeah that's the problem. 20180523 00:07:49<+discordbot> whatever you want to call "+" 20180523 00:08:09<+discordbot> why do we not just use keycodes 20180523 00:08:20<+discordbot> What's a key code? 20180523 00:08:22<+discordbot> zoomin as SDLK_PLUS 20180523 00:08:24<+discordbot> Keycode. 20180523 00:08:27< celticminstrel> I think there are good reasons not to use keycodes but I don't remember what they are. 20180523 00:08:37< gfgtdf> i think the best way is to use scancodes but be able to have something like #if KB_LAYOUT = whatever in the default hotkey preferences file. 20180523 00:08:38< celticminstrel> BTW, SDLK_PLUS is the numeric keypad. 20180523 00:08:41< celticminstrel> I think? 20180523 00:08:42<+discordbot> "The scancode identifies the location of a key press and the corresponding SDL_Keycode gives that key press meaning in the context of the current keyboard layout." 20180523 00:08:58<+discordbot> celmin: no, that's SDLK_KP_PLUS 20180523 00:09:24< celticminstrel> Oh okay sure. 20180523 00:09:41<+discordbot> we use keycodes interally 20180523 00:09:46<+discordbot> when it comes to key press callbacks 20180523 00:09:49<+discordbot> such as in GUI2 20180523 00:10:01< celticminstrel> But does shift+= actually produce SDLK_PLUS? I wouldn't expect it to. 20180523 00:10:28<+discordbot> I can test 20180523 00:10:30<+discordbot> if you want 20180523 00:10:42< celticminstrel> Feel free, I guess? 20180523 00:12:51<+discordbot> ftr, we must use keycodes if we want + 20180523 00:12:54<+discordbot> because it's a virtual key 20180523 00:13:03<+discordbot> it has no scancode 20180523 00:13:11< celticminstrel> ... 20180523 00:13:47-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 00:14:55<+discordbot> what? 20180523 00:15:29-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 00:16:02<+discordbot> looks like it does NOT produce + 20180523 00:16:04<+discordbot> er 20180523 00:16:09<+discordbot> SDLK_PLUS 20180523 00:16:10<+discordbot> It would do us all a big favour if you two stopped it with the constant ellipses and whats. 20180523 00:16:25<+discordbot> pressing Left Shift pressing = 20180523 00:16:25< celticminstrel> Yeah, + is not a virtual key 20180523 00:16:31< celticminstrel> It's not a key at all. 20180523 00:16:34<+discordbot> though 20180523 00:16:39<+discordbot> caveat 20180523 00:16:44<+discordbot> I tested in a GUI2 callback 20180523 00:16:59<+discordbot> celmin: "These virtual key values do not have corresponding physical keys" 20180523 00:17:03<+discordbot> quoth the SDLdocs 20180523 00:17:56< celticminstrel> I believe that may be a misleading statement? What I thought that meant is that the corresponding physical key is not fixed and depends on the keyboard layout or something? 20180523 00:18:00<+discordbot> Does it matter that much what the default hotkeys are, surely, since users can assign our own? 20180523 00:18:15<+discordbot> it does 20180523 00:18:21<+discordbot> because 99% of players won't 20180523 00:18:28< irker017> wesnoth/wesnoth:master Charles Dang b011faa244 GUI2/Canvas: merge draw() and render() a AppVeyor: All builds passed 20180523 00:19:09-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 00:19:38<+discordbot> celmin: there's no scancode entry 20180523 00:19:42<+discordbot> at all 20180523 00:20:01< celticminstrel> Obviously. 20180523 00:20:20<+discordbot> so all you can rely on is the keycode 20180523 00:21:19<+discordbot> actually not sure how to test whether it generates the appropriate keycode.. 20180523 00:24:52< celticminstrel> I'm pretty sure you can convert losslessly between keycodes and scancodes? 20180523 00:25:03< celticminstrel> Provided it's all on one computer. 20180523 00:31:14<+discordbot> Wesnoth's audio skips a lot on my laptop. 20180523 00:31:30<+discordbot> It's actually unusable. 20180523 00:33:48<+discordbot> Also, the language menu entries for Japanese and Simplified Chinese while Spanish or English are enabled show up as glyph placeholders instead of the actual CJK character glyphs, but if I actually switch to either of them, they work. 20180523 00:34:08<+discordbot> *stutters, not skips 20180523 00:34:21<+discordbot> Thought it was skipping at first but it's actually just stuttering. 20180523 00:34:28<+discordbot> It stutters every 0.75 seconds approx. 20180523 00:36:27< celticminstrel> I think that's because Vultraz disabled all the fallback fonts to solve some silly issue. 20180523 00:36:32< celticminstrel> "solve" 20180523 00:36:59< celticminstrel> So it works when it's the active language (because the main font is switched in that case) but not when anything else is the active language. 20180523 00:37:43<+discordbot> If it's the fix for the issue where Pango can't use font fallbacks on Windows without causing empty lines to become 0 pixels tall, it's not "silly". 20180523 00:38:01< celticminstrel> I don't remember if that was what it was. 20180523 00:38:20< celticminstrel> But I guess that's not silly, sure... 20180523 00:39:39<+discordbot> that's what it was 20180523 00:39:52<+discordbot> it's a bug that's been in the pango tracker for 11 years with no fix 20180523 00:40:07<+discordbot> Okay, now I started Wesnoth again and it's not stuttering. 20180523 00:40:14< celticminstrel> ... 20180523 00:40:31<+discordbot> celmin: you can switch between scan and key code but i would assume there need to pairs 20180523 00:40:45<+discordbot> you can't get a nonexistent scancode for a keycode 20180523 00:40:53< celticminstrel> Sure. 20180523 00:41:25< celticminstrel> The thing is, everyone has the same set of scancodes (give or take a few). 20180523 00:41:44<+discordbot> but? 20180523 00:41:49<+discordbot> I sense a but 20180523 00:42:10< celticminstrel> They don't necessarily have the same set of keycodes. 20180523 00:42:32<+discordbot> we could always avoid the virtual keys 20180523 00:42:48<+discordbot> there are scancodes sans keycodes but we'll never need them 20180523 00:43:07<+discordbot> stuff like SDL_SCANCODE_NONUSBACKSLASH 20180523 00:43:12<+discordbot> or SDL_SCANCODE_INTERNATIONAL1 20180523 00:44:04<+discordbot> SDLK_PLUS is the only virtual key we might have use of 20180523 00:44:07<+discordbot> but we don't even use + now 20180523 00:44:10<+discordbot> we use = 20180523 00:44:12<+discordbot> for zoom in 20180523 00:44:13<+discordbot> Regarding writing events as Lua, by the way, one thing I was pondering the other day is if it'd make sense to implement a lua= attribute in [event] as a shorthand for a single [lua] code= block (and then forbid the use of ActionWML for that event perhaps). 20180523 00:44:37<+discordbot> @shadowm literally thinking the exact same thing the other day 20180523 00:44:48< celticminstrel> I've thought about that too. 20180523 00:45:17< celticminstrel> Potentially also a formula= if you want to use WFL, though events kinda want to be more procedural so I dunno if that's a good idea. 20180523 00:45:33<+discordbot> Uh. 20180523 00:45:43<+discordbot> Can WFL do anything at all? 20180523 00:45:47<+discordbot> no 20180523 00:45:49<+discordbot> Action-wise, I mean. 20180523 00:45:54<+discordbot> this is not frogatto 20180523 00:46:00< celticminstrel> Technically yes. 20180523 00:46:12<+discordbot> listens to celticminstrel 20180523 00:46:28< celticminstrel> Basically, FormulaAI. 20180523 00:46:41<+discordbot> which i want to remove 20180523 00:46:41<+discordbot> 😛 20180523 00:46:43< celticminstrel> In FormulaAI, the formula evaluates to a list of actions, which are then executed by the engine. 20180523 00:47:03<+discordbot> But that API is probably far more restricted than what Lua/WML can currently do. 20180523 00:47:06< celticminstrel> Each action is a special formula_callable type. 20180523 00:47:16< celticminstrel> Yeah, the FormulaAI API is indeed fairly restricted. 20180523 00:47:21<+discordbot> e.g. I don't think you can call wesnoth.color_adjust from WFL. 20180523 00:47:52< celticminstrel> You'd need lots more WFL functions if you wanted to extend the FormulaAI mechanism to general events, but it could be done. 20180523 00:48:02< celticminstrel> Is it worth it? I dunno. 20180523 00:48:09<+discordbot> Probably not. 20180523 00:48:19<+discordbot> We already went down the Lua rabbit hole. 20180523 00:48:23< celticminstrel> I did actually generalize the set_var() action to non-AI events. 20180523 00:48:29< celticminstrel> I think that's used in like, one place. 20180523 00:48:33<+discordbot> no, it's not worth it 20180523 00:48:33<+discordbot> We're not getting out of wonderland any time soon. 20180523 00:48:44<+discordbot> I considered it as a possibility 20180523 00:48:46<+discordbot> since ya know 20180523 00:48:48<+discordbot> frogatto does it 20180523 00:48:52<+discordbot> but I decided against it 20180523 00:48:59<+discordbot> which is why I want to remove FAI so we can focus on Lua. 20180523 00:49:08< celticminstrel> Fair enough. 20180523 00:49:37<+discordbot> frogatto also has FSON besides just FFL 20180523 00:49:40<+discordbot> which is integral 20180523 00:49:47<+discordbot> (FSON, not JSON) 20180523 00:49:49<+discordbot> I'm going to translate for the audience. 20180523 00:50:11<+discordbot> Anura (the engine Frogatto uses) started out using a subset of WML as its container language. 20180523 00:50:35<+discordbot> With a variation on the SilverTree Formula Language as its scripting language. 20180523 00:50:46<+discordbot> (The SilverTree Formula Language is the direct ancestor of WFL.) 20180523 00:51:12<+discordbot> Eventually, they replaced that WML subset with a JSON-like language. 20180523 00:51:30<+discordbot> The formula language continued to evolve far beyond what WFL has to offer. 20180523 00:51:56<+discordbot> I think it's actually not just a functional programming language anymore. 20180523 00:52:24<+discordbot> IIRC it's gained things like object prototypes and strict typing and other stuff. 20180523 00:52:30< celticminstrel> My understanding was that it uses something very similar to what I described in FormulaAI, with formulas that return a list of actions to be executed by the engine? 20180523 00:52:34<+discordbot> strict typing, yes 20180523 00:52:45< celticminstrel> Yeah I noticed the strict typing, too. 20180523 00:53:00<+discordbot> object prototypes are in FSON 20180523 00:53:01<+discordbot> Their JSON-like language is to their formula language what WML is to Lua in Wesnoth. 20180523 00:53:49<+discordbot> I'm just being told they made some improvements compared to JSON 20180523 00:54:11<+discordbot> Basically it'd be like if we got rid of WML actions entirely and replaced event scripting in WML with Lua. 20180523 00:54:41<+discordbot> Relegating WML back to what it really should've been used for in the first place, describing things rather than expressing executable code. 20180523 00:56:17< celticminstrel> I'd be totally fine with doing that in some hypothetical 1.22 or so. 20180523 00:56:37<+discordbot> we'll be dropping the 1.x versioning at somepoint 20180523 00:57:15< celticminstrel> I mean, it's true that the 1.x versioning is totally useless, but just dropping it causes compatibility issues, so you need to be careful. 20180523 00:57:32<+discordbot> it'd perhaps be in another stable series 20180523 00:57:44< celticminstrel> ... 20180523 00:57:54<+discordbot> My original plan was to do it in 1.13.x. 20180523 00:58:37<+discordbot> I didn't tell anyone I was going to do it because I knew it'd become a controversial topic. 20180523 00:59:06<+discordbot> Also what did I say about the whats and ellipses? 20180523 01:02:51-!- gfgtdf [~gfgtdf@134.76.63.8] has quit [Quit: Leaving] 20180523 01:45:38<+discordbot> celmin: how would one run the contents of a lua= key from c++? 20180523 01:47:06<+discordbot> wait, hang on 20180523 01:48:36<+discordbot> handle_event_commands 20180523 01:48:38<+discordbot> in lua 20180523 01:50:36< irker017> wesnoth: Charles Dang wesnoth:master a1de152dedc2 / src/game_events/manager.cpp: Game Events/Manager: formatting cleanup https://github.com/wesnoth/wesnoth/commit/a1de152dedc2f83a61534714a319c01df092f0c4 20180523 01:50:59<+discordbot> or maybe in command... 20180523 01:51:04<+discordbot> celmin: what say you? 20180523 01:52:16< irker017> wesnoth/wesnoth:master ancestral 6b2b324fe4 Fixed speak commands AppVeyor: All builds passed 20180523 01:59:01<+discordbot> lua function wml_actions.command(cfg) if cfg.lua then wml_actions.lua { code = cfg.lua } else utils.handle_event_commands(cfg, "plain") end end 20180523 01:59:56<+discordbot> testcase (works): [event] name=start lua = << wesnoth.confirm('foo', 'bar') >> [/event] 20180523 01:59:57<+discordbot> testcase (works): [event] name=start lua = << wesnoth.confirm('foo', 'bar') >> [/event] 20180523 02:07:30<+discordbot> It could be code too, but I thought lua made more sense since it's the tag's name. 20180523 02:07:43<+discordbot> indeed 20180523 02:08:03<+discordbot> probably should also have something like this: lua if #wml.shallow_literal(cfg) > 1 then helper.wml_error("ActionWML not accepted when using lua= key") end 20180523 02:08:07<+discordbot> if that's how one does it 20180523 02:08:36<+discordbot> Well... 20180523 02:08:43<+discordbot> The filter tags would still be accepted. 20180523 02:08:48<+discordbot> Right? 20180523 02:10:27<+discordbot> Okay so with keys 20180523 02:10:34<+discordbot> ...ah 20180523 02:10:38<+discordbot> 🤔 20180523 02:10:48<+discordbot> perhaps this should be in handle_event_commands after all 20180523 02:11:24< Ravana_> that is just to allow you avoid writing [lua][/lua]? 20180523 02:11:33<+discordbot> yes 20180523 02:12:27<+discordbot> maybe this is too much of special syntax case 20180523 02:12:55<+discordbot> If I have [hotkey] command=continue key="space" alt=no [/hotkey] and later [hotkey] command=endturn key="space" alt=yes [/hotkey] Will this conflict? Because if I leave alt=no out it certainly will. 20180523 02:12:57<+discordbot> plus it means one doesn't have the other features of [lua] 20180523 02:13:00<+discordbot> like [args] 20180523 02:13:26<+discordbot> Not sure if [args] makes much sense when you're already using Lua. 20180523 02:13:48<+discordbot> It's normally only used to pass WML to it. 20180523 02:13:52< Ravana_> I tend to define new wml tag instead of using direct [lua] in nonpreload events now 20180523 02:13:57<+discordbot> I've never used it but it certainly could be useful in a few corner cases 20180523 02:14:02<+discordbot> I have. 20180523 02:15:44<+discordbot> kinda leaning towards just keeping [lua] 20180523 02:15:47<+discordbot> for consistency 20180523 02:16:08<+discordbot> I'd also have to add a conditional lua= key if i add this 20180523 02:16:16<+discordbot> ([lua] is also a conditional action now) 20180523 02:16:36<+discordbot> Is it? 20180523 02:16:44<+discordbot> yes 20180523 02:16:57<+discordbot> late api addition 20180523 02:17:05< Ravana_> few weeks ago it was decided against having inline conditional Lua 20180523 02:17:30< Ravana_> related to lua_function= in filters 20180523 02:18:12<+discordbot> lua_function is dumb. We should allow inline conditional lua 20180523 02:18:44<+discordbot> especially since one can just do [lua] code = wesnoth.eval_conditional [/lua] anyway 20180523 02:19:58-!- travis-ci [~travis-ci@ec2-54-162-4-12.compute-1.amazonaws.com] has joined #wesnoth-dev 20180523 02:19:59< travis-ci> wesnoth/wesnoth#18325 (master - a1de152 : Charles Dang): The build was broken. 20180523 02:19:59< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/382460593 20180523 02:19:59-!- travis-ci [~travis-ci@ec2-54-162-4-12.compute-1.amazonaws.com] has left #wesnoth-dev [] 20180523 02:20:25<+discordbot> wat 20180523 02:21:06<+discordbot> ah 20180523 02:21:24<+discordbot> "Could not initialize SDL_video error, retrying..." so not a problem with my code 20180523 02:22:07<+discordbot> celmin: I move to deprecate lua_function 20180523 02:32:13<+discordbot> huh 20180523 02:32:14<+discordbot> odd 20180523 02:32:30<+discordbot> images aren't opening on that build 20180523 02:32:31<+discordbot> oh well 20180523 02:32:49< irker017> wesnoth: ancestral wesnoth:cleanup_keys 7616cd3b5fc7 / data/core/hotkeys.cfg: Update hotkeys.cfg https://github.com/wesnoth/wesnoth/commit/7616cd3b5fc73eb12773f7e640712de6f2c3ce83 20180523 02:32:59<+discordbot> all builds, actually... 20180523 02:33:02<+discordbot> very odd! 20180523 02:33:15<+discordbot> I’m done 20180523 02:58:44-!- travis-ci [~travis-ci@ec2-54-166-169-32.compute-1.amazonaws.com] has joined #wesnoth-dev 20180523 02:58:45< travis-ci> wesnoth/wesnoth#18326 (cleanup_keys - 7616cd3 : ancestral): The build was broken. 20180523 02:58:45< travis-ci> Build details : https://travis-ci.org/wesnoth/wesnoth/builds/382469661 20180523 02:58:45-!- travis-ci [~travis-ci@ec2-54-166-169-32.compute-1.amazonaws.com] has left #wesnoth-dev [] 20180523 03:04:33< celticminstrel> @Vultraz Okay so what do you want exactly? 20180523 03:04:47<+discordbot> to deprecate lua_function 20180523 03:05:17<+discordbot> I think it's best not to add a lua= tag under [command] (and by extension, [event], so nevermind that 20180523 03:05:49< celticminstrel> @ancestral Why would space and alt+space conflict? 20180523 03:06:28< celticminstrel> Okay. 20180523 03:06:32<+discordbot> but yeah, I propose we deprecate lua_function 20180523 03:06:35< celticminstrel> Hm. 20180523 03:06:41<+discordbot> perhaps in favor of an inline lua= key 20180523 03:07:10< celticminstrel> Well, I would definitely prefer to write the Lua inline in the filter, rather than referencing a function defined elsewhere. 20180523 03:07:12<+discordbot> but in filters only 20180523 03:07:28<+discordbot> celticminstrel: If space comes before alt-space, alt-space will never trigger. 20180523 03:07:35<+discordbot> or we could drop that entirely and make people use [lua] wesnoth.eval conditinal 20180523 03:07:37<+discordbot> AFAICT 20180523 03:07:39< celticminstrel> To satisfy gfgtdf you probably need to make sure the code string is parsed only once per filter. 20180523 03:08:07< celticminstrel> @Vultraz, ActionWML and ConditionalWML don't really compare to FilterWML. 20180523 03:08:10<+discordbot> @ancestral well, yes. space alt is not alt space 20180523 03:08:19< celticminstrel> The latter already supports an inline formula, for example, so adding inline Lua makes a lot of sense. 20180523 03:08:35<+discordbot> ok 20180523 03:08:45<+discordbot> Not sure you understand 20180523 03:08:47< celticminstrel> And having inline Lua in FilterWML but not in Action/ConditionalWML doesn't seem like a problem, as the latter also don't have inline formulas. 20180523 03:08:56<+discordbot> indeed 20180523 03:09:00<+discordbot> that's what I was thinking 20180523 03:09:04< celticminstrel> @ancestral Why won't it trigger? 20180523 03:09:17<+discordbot> If I have a key defined for space, and later in the file I have one with alt+space, it doesn’t seem to trigger the second one. 20180523 03:09:28<+discordbot> Don’t know why. I’m not much of a C++ programmer. 20180523 03:09:29<+discordbot> .......... 20180523 03:09:32< celticminstrel> That seems like it could be a bug... 20180523 03:09:33<+discordbot> (╯°□°）╯︵ ┻━┻ 20180523 03:09:49<+discordbot> I mean, it makes sense that it finds the first hotkey and runs with it 20180523 03:09:59<+discordbot> ONE MILLION HELLS 20180523 03:10:05< celticminstrel> No, no, no. 20180523 03:10:28< celticminstrel> A key with alt=no and a key with alt=yes have no relation, it should be impossible to match the one with alt=no when the alt key is pressed. 20180523 03:10:32<+discordbot> Unless maybe I have alt=no, but it seems wasteful 20180523 03:10:39<+discordbot> to have that in most cases 20180523 03:10:49< celticminstrel> @ancestral Can you please explain the purpose of 3157, I don't get it 20180523 03:11:01<+discordbot> simpler hotkeys 20180523 03:11:08<+discordbot> The first was just re-orging the file 20180523 03:11:16<+discordbot> The second was new keys and deprecating a couple 20180523 03:11:31<+discordbot> I wanted to do it in two edits so you wouldn’t be like WTF just happened. 20180523 03:11:43< celticminstrel> What's the goal of the reorg? 20180523 03:11:56<+discordbot> The aforementioned “bug” (apparently) 20180523 03:12:02< celticminstrel> No, but really. 20180523 03:12:36<+discordbot> If I define a key before a command+key, the command+key doesn’t fire. 20180523 03:12:46<+discordbot> how nice... 20180523 03:12:50<+discordbot> Unless, maybe I put ctrl=no? 20180523 03:13:02<+discordbot> Well, command=no and ctrl=no 20180523 03:13:05<+discordbot> do you really need to reorganize anything 20180523 03:13:30<+discordbot> It’s alphabetical by section 20180523 03:13:41<+discordbot> The editor commands are still together 20180523 03:13:48<+discordbot> The main menu commands are still together 20180523 03:14:16<+discordbot> The game commands are in two sections: Command/ctrl and non-Command/ctrl, and each of those are alphabetical 20180523 03:14:26<+discordbot> Otherwise, I would have a huge section that cannot be alphabetical 20180523 03:15:45<+discordbot> (I might have one or two that are out of order to get around the modifier key ordering issue I already mentioned, to avoid having a third section.) 20180523 03:16:34<+discordbot> I also enabled delay shroud hotkey 20180523 03:18:57<+discordbot> Leader becomes K Planning mode becomes Alt+P Suppose Dead becomes X Undo becomes Z and Ctrl/Cmd+Z Redo becomes Shift Z and Ctrl/Cmd+Z (if you prefer Y here, I could make that work) Delete Action becomes Backspace Continue becomes Space 20180523 03:19:08< celticminstrel> So alphabetical by command id? 20180523 03:19:09<+discordbot> Many other common commands gain an extra key without deprecating the original. 20180523 03:19:15<+discordbot> Yes 20180523 03:19:31< celticminstrel> And secondarily sorted by modifier set? 20180523 03:20:16< celticminstrel> When you said "modifiers appear before non-modifiers" I thought maybe you meant the order of attributes in each hotkey definition. 20180523 03:20:21<+discordbot> 1. Command/Control Modifier Hotkeys 2. Game Hotkeys 3. Editor Hotkeys 4. Main Menu Hotkeys 20180523 03:20:27<+discordbot> Each section is sorted alphabetically by command 20180523 03:20:42< celticminstrel> I don't really get section 1 but okay. 20180523 03:21:23< celticminstrel> I guess you're only doing the hotkeys that already have defaults, so nothing for start test scenario. 20180523 03:21:34<+discordbot> celticminstrel: If I define a shortcut R and another Ctrl+R, if the R is first int he file, the Ctrl+R will never trigger when pushed. 20180523 03:21:45< celticminstrel> Sounds like a bug. 20180523 03:22:09< celticminstrel> Does it work if you assign the keys in the UI? 20180523 03:22:18< celticminstrel> Even after relaunching the game? 20180523 03:22:26< celticminstrel> No matter which order you assigned them in? 20180523 03:22:34<+discordbot> Not sure' 20180523 03:22:35< celticminstrel> Did adding explicit alt=no or whatever actually help? 20180523 03:23:40<+discordbot> I haven’t tried 20180523 03:23:47<+discordbot> But I was tempted to 20180523 03:23:57< celticminstrel> Could you try? 20180523 03:24:06<+discordbot> Yeah 20180523 03:24:35<+discordbot> Still, for a command and ctrl key 20180523 03:24:42<+discordbot> That means both of those would be no? 20180523 03:24:51< celticminstrel> Uhh. 20180523 03:24:52<+discordbot> How do you handle Apple command = no 20180523 03:25:05<+discordbot> Because the key “command” is sorta taken 20180523 03:25:09< celticminstrel> I think it would be fine as long as you set them both to no before using the macro? 20180523 03:25:12<+discordbot> Oh, cmd 20180523 03:25:13<+discordbot> Sorry 20180523 03:25:28< celticminstrel> Then the key added by the macro overrides the previous explicit key. 20180523 03:25:31< celticminstrel> I think. 20180523 03:25:41< celticminstrel> Pretty sure that's how WML works in the case of duplicate keys. 20180523 03:28:24< celticminstrel> Hmm, looking at src/hotkey/hotkey_item.cpp I don't see any obvious cause for this bug... 20180523 03:28:50< celticminstrel> AFAICT the mod bitfield is always compared with equality to the actual modifier state from SDL. 20180523 03:31:38<+discordbot> celticminstrel: Did not work 20180523 03:31:55<+discordbot> Wait 20180523 03:32:00<+discordbot> Sorry, maybe it did 20180523 03:33:29<+discordbot> Hmm 20180523 03:34:44<+discordbot> Well now I can’t seem to reproduce the bug 20180523 03:35:00< celticminstrel> :( 20180523 03:35:13<+discordbot> So either I am assuming there’s an issue when there’s not, or there is and it’s very conditional 20180523 04:23:54< irker017> wesnoth/wesnoth:1.14 Severin Glöckner 2e04dde374 updated German translation AppVeyor: All builds passed 20180523 04:27:04-!- celticminstrel [~celmin@unaffiliated/celticminstrel] has quit [Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] 20180523 04:57:43-!- boucman [~rosen@wesnoth/developer/boucman] has joined #wesnoth-dev 20180523 05:22:46<+discordbot> After further analysis, it looks like the memory leak is related to reach maps. 20180523 05:23:09<+discordbot> Wesnoth leaks a couple of megabytes every time I hover a Silver Mage in that map. 20180523 05:24:02<+discordbot> Due to teleport ability, the sheer size of the map, and the number of towns, the Silver Mage can reach whopping 5787 hexes. 20180523 05:25:13<+discordbot> Most likely there is some kind of leak every time any unit is hovered, but the size of the leak is related to the number of reachable hexes. 20180523 05:42:31-!- boucman [~rosen@wesnoth/developer/boucman] has quit [Remote host closed the connection] 20180523 06:06:13< irker017> wesnoth/wesnoth:cleanup_keys ancestral ccc223b30f Reorganized hotkeys file so modifiers ap AppVeyor: All builds passed 20180523 06:34:10<+discordbot> @jyrkive which component therein? 20180523 06:34:58<+discordbot> Are you asking "which part of reach map handling leaks memory?" 20180523 06:35:04<+discordbot> If so, I don't know. 20180523 06:35:10<+discordbot> there's a display::reach_map object in display and a pathfind::marked_route object in game_display 20180523 06:35:45<+discordbot> My guess is the former. 20180523 06:35:58<+discordbot> typedef std::map reach_map; 20180523 06:36:25<+discordbot> hmmmm 20180523 06:37:12<+discordbot> Remember that in addition to storing the map locations (they're just eight bytes each: even 6000 map location objects takes only 48 kB), there is also reachmap rendering code. 20180523 06:37:29<+discordbot> ah, true 20180523 06:37:42<+discordbot> That's still what I'd suspect first. Leaking images can leak memory much faster. 20180523 06:37:49<+discordbot> which is significantly different in master.. I should be looking at 1.14 20180523 06:38:49<+discordbot> The next debugging tehnique I was planning to try is disabling some parts of reachmap handling. 20180523 06:39:06<+discordbot> If disabling X fixes the leak, then X is responsible. 20180523 06:40:36<+discordbot> there's also a pathfind::marked_route and pathfind::paths in events::mouse_handler... 20180523 06:51:10-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has joined #wesnoth-dev 20180523 09:09:02-!- irker017 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout] 20180523 09:11:20-!- irker756 [~irker@uruz.ai0867.net] has joined #wesnoth-dev 20180523 09:11:20< irker756> wesnoth/wesnoth:1.14 gfgtdf d1097d93d5 fix oos when using 'back to turn' in mp AppVeyor: All builds passed 20180523 10:11:37-!- atarocch [~atarocch@93.56.164.28] has quit [Quit: Leaving] 20180523 10:11:53-!- atarocch [~atarocch@93.56.164.28] has joined #wesnoth-dev 20180523 10:16:04-!- atarocch [~atarocch@93.56.164.28] has quit [Client Quit] 20180523 10:16:21-!- valdar [~atarocch@93.56.164.28] has joined #wesnoth-dev 20180523 10:46:25< irker756> wesnoth/wesnoth:master Charles Dang a1de152ded Game Events/Manager: formatting cleanup AppVeyor: All builds passed 20180523 10:56:01< irker756> wesnoth: gfgtdf wesnoth:1.14 8419db8a452f / src/playsingle_controller.cpp: don't autosave after using 'back to turn' in mp https://github.com/wesnoth/wesnoth/commit/8419db8a452fae7c9e25e211cc876aee83c5a2b1 20180523 11:02:35< irker756> wesnoth: gfgtdf wesnoth:master 31d4579500cb / src/playmp_controller.cpp: fix oos when using 'back to turn' in mp https://github.com/wesnoth/wesnoth/commit/31d4579500cb840fb8de36fbffdb8a197ae6e4f7 20180523 11:02:37< irker756> wesnoth: gfgtdf wesnoth:master fb54b0376d0f / src/playsingle_controller.cpp: don't autosave after using 'back to turn' in mp https://github.com/wesnoth/wesnoth/commit/fb54b0376d0fff9250d8947e269587cc60df6a6c 20180523 11:24:43<+discordbot> Bug Report: In 1.14, the game assume the uniqueness of "canrecruit=yes". If you create a second such unit with [unit], the tag would do nothing. 20180523 11:26:03<+discordbot> An example can be the first scenario of "The Final Exam" UMC campaign. 20180523 11:27:54<+discordbot> hmm... 20180523 11:28:00<+discordbot> very odd 20180523 12:23:50-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection] 20180523 12:24:03-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth-dev 20180523 12:50:32<+discordbot> @Denivarius @Vultraz @shadowm https://forums.wesnoth.org/viewtopic.php?f=2&t=48235 20180523 12:51:13<+discordbot> Denivarius is Dave right? 20180523 12:51:51<+discordbot> Yes, he is. 20180523 12:56:17<+discordbot> is this the scheme google follows with chrome? 20180523 12:56:53<+discordbot> chromium version seems to have 4 components 20180523 12:57:29<+discordbot> ah, yes 20180523 12:57:35<+discordbot> https://www.chromium.org/developers/version-numbers 20180523 12:57:42<+discordbot> It's incompatible with semver. 20180523 13:03:09<+discordbot> only downside I see is that now version numbers would mean a lot more than they do now 20180523 13:08:32<+discordbot> Not really my gig so I'll shut up after this comment, but surely the MINOR/PATCH level distinctions are just as important in the dev release. Having all the dev releases between major versions be called 15.0.0-alpha.(X) seems a bit clumsy. 20180523 13:18:19-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 13:20:04<+discordbot> I agree 20180523 13:23:14<+discordbot> Well I like it more than having dev releases deceptively look like stable ones 20180523 13:23:28<+discordbot> unless you know about even/odd convention 20180523 13:23:32<+discordbot> which most people don't 20180523 13:24:00<+discordbot> And I don't see how MINOR/PATCH would be useful for dev releases 20180523 13:24:39<+discordbot> Agreed. It's bad that nothing distinguishes stable and unstable releases, other than the second number being odd/even. 20180523 13:26:04< irker756> wesnoth/wesnoth:master ancestral 37138852d7 Update hotkeys.cfg AppVeyor: All builds passed 20180523 13:28:05<+discordbot> That is bad. But it is simple and is explained everywhere that people can download the dev version. 20180523 13:30:08<+discordbot> Not sure how to resolve this 20180523 13:41:31-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 13:47:38<+discordbot> in any case I'd want to know what shadowm, celmin, zookeeper etc think too 😛 20180523 13:55:36-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 14:20:15-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 245 seconds] 20180523 14:23:23-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 15:09:37<+discordbot> How is the odd versioning for dev releases any less compatible with the semver than with the current versioning? 20180523 15:12:39-!- gfgtdf [~gfgtdf@134.76.63.8] has joined #wesnoth-dev 20180523 15:12:41-!- gfgtdf_ [~gfgtdf@134.76.63.8] has joined #wesnoth-dev 20180523 15:12:59-!- gfgtdf_ [~gfgtdf@134.76.63.8] has quit [Client Quit] 20180523 15:13:55< gfgtdf> i wonder why i don't see this thread https://forums.wesnoth.org/viewtopic.php?f=2&p=628670#p628670 when i go to this page https://forums.wesnoth.org/search.php?search_id=active_topics ? 20180523 15:14:59< gfgtdf> it does appear on this page though https://forums.wesnoth.org/app.php/feed (which i used to use but somehow my mobile browser cannot display it well) 20180523 15:18:04-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 256 seconds] 20180523 15:37:24<+discordbot> @Bitron Semver requires that API-incompatible changes always increment the major version number. 20180523 15:38:23<+discordbot> Thus, with it we'd end up having to increase the major version number for every development release. For example, 1.13.0, 1.13.1, 1.13.2 and 1.13.3 would instead have to be 1.13.0, 2.0, 3.0 and 4.0. 20180523 15:39:03<+discordbot> (And since we maintain API compatibility with stable releases, stable release branches would not increment the major version number in every release.) 20180523 15:51:53<+discordbot> Oh. That makes sense, then. What about giving dev releases four segments? Means devver.major.minor.bugfix So that next dev release would be 15.1.0.0. (if I'm not mistaken) 20180523 15:52:22<+discordbot> Well, that's not semantic versioninc. Semver is extremely strict about what it means. 20180523 15:52:47<+discordbot> I see. 20180523 15:57:37<+discordbot> what's the upside of forcing everything into fitting semver, then? 20180523 16:01:18<+discordbot> I don't know. 20180523 16:13:36-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection] 20180523 16:13:43-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth-dev 20180523 16:19:37<+discordbot> @loonycyborg I don't understand why you want Dave's opinion. 20180523 16:20:07<+discordbot> If it you were tagging past release managers then I find it interesting that you forgot about our longest-running release manager, Ivanovic. 20180523 16:21:09<+discordbot> well I think dave would be the one who came up with versioning scheme in the first place, not sure though 20180523 16:21:40<+discordbot> I think you could replace our version numbers with species of tropical fish and he wouldn't be particularly concerned either way. 20180523 16:22:15<+discordbot> 15.0.0-barracuda? 😛 20180523 16:22:23<+discordbot> No, just the fish. 20180523 16:22:29<+discordbot> No numbers. 20180523 16:24:26<+discordbot> This is more of a "topic" for people who actually use Wesnoth (me and other content developers and players) and people who develop Wesnoth (insert list of active developers here), really. 20180523 16:25:35<+discordbot> Anyway your proposal is flawed because it's not just a number format change proposal, it's a workflow change proposal. 20180523 16:26:23-!- irker756 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout] 20180523 16:26:52<+discordbot> I'm not a fan of using hyphens in version numbers either. 20180523 16:27:12<+discordbot> Or any non-numeric characters at all. 20180523 16:28:06<+discordbot> well actually it doesn't change workflow by itself 20180523 16:28:11<+discordbot> There's a whole unit test dedicated to enforcing version_info's semantics and deviating from that in one go doesn't sound like the best idea to me as it'll confuse everyone. 20180523 16:28:36<+discordbot> The premise of point #5 is that there's been a huge change in the release cycle for stable versions. 20180523 16:29:16<+discordbot> good point, but we aren't forced to increment minor version 20180523 16:29:27<+discordbot> We've discussed extrensively why it's not a good idea to make API changes in stable. 20180523 16:29:30<+discordbot> and related workflow change can come at its own pace 20180523 16:29:56<+discordbot> It'd be good if you made sure to include this information in the proposal then. 20180523 16:29:57<+discordbot> that applies to forward compatible changes too? 20180523 16:31:17<+discordbot> I think the distinction is a bit hard to make in here, especially when everyone has their own idea about how to write or deploy their code. 20180523 16:31:40<+discordbot> this mainly applies to umc though 20180523 16:31:52<+discordbot> semantic versioning is primarily concerned with api 20180523 16:32:04<+discordbot> and in this case our api is to umc 20180523 16:32:51<+discordbot> And that's what I'm talking about. 20180523 16:34:07<+discordbot> So distinction amounts to "doesn't break existing umc" 20180523 16:49:03-!- boucman [~rosen@wesnoth/developer/boucman] has joined #wesnoth-dev 20180523 17:04:29< Ivanovic> @loonycyborg define "does not break"... 20180523 17:04:39< Ivanovic> almost any change could be seen as a possible break 20180523 17:04:49< Ivanovic> of course any balancing change to unit stats 20180523 17:05:03< Ivanovic> but as well changes to the AI since it is now "smarter" and needs a rebalance of maps 20180523 17:05:19< Ivanovic> and not to mention changes to the lore which could impact how UMC fits into the existing lore 20180523 17:05:50< Ivanovic> following that theory we should not have any dev releases against which someone might develop UMC since we already know that we are "free" to change anything until we are about to start the next stable series 20180523 17:06:19<+discordbot> break means only ability to run it 20180523 17:06:53< Ivanovic> if you are talking about API in the strictest sense then any change we do has the impact to break UMC and would require a major version update 20180523 17:06:54<+discordbot> I don't expect balance changes in umc due to changes to mainline units will be too great 20180523 17:06:57-!- Bhoren [~Bhoren_wh@2a01:e0a:c:2150:d19d:a3c:7917:5a2d] has joined #wesnoth-dev 20180523 17:06:59< Ivanovic> okay, not typo fixes 20180523 17:07:16< Ivanovic> not too great != does not have the potential to break 20180523 17:07:43<+discordbot> potential yes, but I don't think we should cover it with version numbers 20180523 17:08:06<+discordbot> it's strictly about interaction between software components 20180523 17:08:18<+discordbot> like which umc tags it's using 20180523 17:10:04<+discordbot> And major unit rebalance would probably need a major version increment in any case 20180523 17:10:34<+discordbot> since you're not forbidden from incrementing it just in response to large scale changes that don't change api 20180523 17:12:19< Ivanovic> again, what is the benefit of this change? 20180523 17:12:34< Ivanovic> since it would assume that we keep running a UMC server for each of those "major versions" 20180523 17:12:58< Ivanovic> otherwise, if they shared all the same server until we reached a stable release again where it branches off: no benefit?! 20180523 17:13:17< Ivanovic> and setting up 10 different UMC servers to run for ages is no benefit as well 20180523 17:13:27< Ivanovic> yes, in some aspects Wesnoth can be seen as API 20180523 17:13:40< Ivanovic> in other aspects it is still "just a game" that people create mods for 20180523 17:14:45< Ivanovic> actually it would encourage a lot more "unexperienced" users to try the non-stable version 20180523 17:15:11< Ivanovic> which might be very bad feedback for us unless we made sure that each of the dev releases is more stable than the ones in the past often were 20180523 17:16:19< Ivanovic> (if it is just "one number higher" people will easily accept it, but once you reach like 10 numbers difference, there is more of a psychological pressure to upgrade because you are so far behind) 20180523 17:17:22<+discordbot> only change is phasing out vestigial 1. and track dev releases like x.0.0-alpha.y 20180523 17:17:32<+discordbot> and have a minor version too 20180523 17:17:51<+discordbot> I think such notation is needed for dev releases 20180523 17:18:05<+discordbot> so people are less likely to package them in stable version of distros 😛 20180523 17:18:26<+discordbot> like fedora taking 1.13 releases 20180523 17:18:35<+discordbot> Don't call such changes semantic versioning, then. That's not semver. 20180523 17:18:48<+discordbot> it makes them compliant to semver 20180523 17:19:10<+discordbot> in fact semver is merely accumulated experience of people versioning their software 20180523 17:19:19<+discordbot> I merely take advantage of it 😛 20180523 17:19:47<+discordbot> No, it's not compliant. 20180523 17:20:06<+discordbot> Semver doesn't allow incrementing the minor version when there aren't any API additions. 20180523 17:21:03<+discordbot> If you ask me, semver is so restrictive that software projects should not adopt it. 20180523 17:21:25<+discordbot> (Nothing blocks use of semver-inspired versioning, however.) 20180523 17:22:02<+discordbot> "It MAY be incremented if substantial new functionality or improvements are introduced within the private code." 20180523 17:22:13<+discordbot> it is minor version here 20180523 17:24:32< Ivanovic> 1) I dislike the text part in there because it makes it a hell of a lot more difficult to do "decent sorting" 20180523 17:25:03< Ivanovic> 2) so your goal is to make sure that distros that act stupid reconsider? 20180523 17:25:16< Ivanovic> why not send them a mail if we know they are stupid and tell them so... 20180523 17:25:56< Ivanovic> 3) I consider those "elements" like an "alpha" marking to be more or less build specific 20180523 17:26:07< Ivanovic> for us this would apply for specific binaries but not for the source release 20180523 17:26:25<+discordbot> You did tag a version 1.6a. 20180523 17:26:28< Ivanovic> yes 20180523 17:26:39< Ivanovic> which should better have been a 1.6.1 20180523 17:26:42<+discordbot> (sic, since it's not 1.6.0a.) 20180523 17:27:02<+discordbot> (Even to this day the fact that the branch and the tag are both called 1.6 causes me some grief when I need to review history.) 20180523 17:27:44<+discordbot> (We probably should've asked for that to be changed when converting from svn.) 20180523 17:28:51<+discordbot> I think -alpha stuff is the most hard to implement part here 20180523 17:29:01<+discordbot> I might be open to changing it 20180523 17:29:14<+discordbot> maybe even to stick with old odd/even convention 20180523 17:29:20<+discordbot> but I don't like it much 20180523 17:31:06< Ivanovic> I think for all the problems with some distros acting "strange" it is really cool that we have the steam release 20180523 17:31:21< Ivanovic> we can just keep that at "stable" and deliver dev binaries there as some beta channel 20180523 17:32:10< Ivanovic> this will probably already solve a lot of the "problems" that less experienced users have, no matter what the distro ships or does not ship since we can always point there 20180523 17:32:42< Ivanovic> and if someone asks why the stable version is not supported by their distro or they include a dev version, we can redirect them to the packager to ask them to read our documentation on versions 20180523 17:33:00< Ivanovic> okay, we could rename "dev" to "unstable" 20180523 17:33:00<+discordbot> still I think my proposed scheme would better reflect our current process too 20180523 17:33:25<+discordbot> like it will have same major number for both dev, rc and stable 20180523 17:33:37<+discordbot> which makes sense since they originate from same continuity 20180523 17:33:41<+discordbot> from same branch 20180523 17:34:56<+discordbot> this might be a good payoff for supporting weirder notations like -alpha and -rc 20180523 17:35:48<+discordbot> You know about version_info, right? 20180523 17:36:16<+discordbot> Although in hindsight I should've called it version_number. 20180523 17:38:35<+discordbot> I know only about src/wesconfig.h 20180523 17:38:42<+discordbot> where does version_info live? 20180523 17:39:00<+discordbot> In src/version.hpp. 20180523 17:40:18<+discordbot> The game version defined there is stored as a version_info and the semantics of version_info are exposed through the #ifver WML preprocessor directive and the wesnoth.compare_versions Lua function. 20180523 17:40:51<+discordbot> You can read src/tests/test_version.cpp to get an idea of what it works like. 20180523 17:41:12<+discordbot> I'm closing in on the Silver Mage memory leak. And listening fitting music. 🎵 20180523 17:41:13<+discordbot> https://www.youtube.com/watch?v=u_dgUf8VHd8 20180523 17:46:26<+discordbot> version_info supports things like "1.5.10-1.6beta2" 20180523 17:46:48<+discordbot> I guess it will support "15.0.0-rc1" too 20180523 17:47:04<+discordbot> not sure if it will support "15.0.0.-rc.1" 20180523 17:47:18<+discordbot> Not in the way you probably expect. 20180523 17:47:52<+discordbot> "15.0.0.-rc.1" is { "15.0.0.0", '-', rc.1 } 20180523 17:48:35<+discordbot> "15.0.0.0" < "15.0.0.-rc.1" holds true. 20180523 17:48:51<+discordbot> (Yes, I'm carrying your trailing dot around because I have no idea what's up with it.) 20180523 17:49:01-!- irker286 [~irker@uruz.ai0867.net] has joined #wesnoth-dev 20180523 17:49:01< irker286> wesnoth/wesnoth:1.14 gfgtdf 8419db8a45 don't autosave after using 'back to turn AppVeyor: All builds passed 20180523 17:49:22<+discordbot> to make parsing out this number easier I guess 20180523 17:49:23<+discordbot> For that matter, "1.15.0" < "1.15.0-dev" also holds true. 20180523 17:50:04<+discordbot> I believe when I wrote the patch (which was mostly aimed at UMC) it was the first development cycle (1.5.x) that had a -dev "version" before the first release. 20180523 17:50:22<+discordbot> The previous cycle (1.3.x.) had its first release numbered 1.3.1. 20180523 17:50:52<+discordbot> oh 20180523 17:50:55<+discordbot> I believe the "version" before 1.3.1 was 1.3.0+dev despite the fact that there wasn't a 1.3.0. 20180523 17:50:57<+discordbot> trailing dot was typo 20180523 17:51:28<+discordbot> Oh, actually it was just 1.3-svn. 20180523 17:52:09<+discordbot> Then the next cycle it became 1.5.0-svn, followed by 1.5.0, followed by 1.5.0+svn (yes, we used svn instead of dev until we migrated to Git). 20180523 17:53:09<+discordbot> I think at some point I considered making a special case for - vs. + but decided against it because it'd make the logic too complicated and like I said, it was originally aimed at UMC. 20180523 17:54:08<+discordbot> Specifically, it was intended to aid the implementation of the update feature in the add-ons management code (I was later informed that I could've used the add-on's upload timestamp instead, but in all honesty that's not necessarily indicative of a real update either). 20180523 17:54:40<+discordbot> At some point later someone decided to piggyback on version_info to do the save version check and thus the game version became a version_info too. 20180523 17:55:02<+discordbot> Much, much later I implemented the #ifver preprocessor directive and the WESNOTH_VERSION macro. 20180523 17:56:07<+discordbot> #ifver is actually general enough that it can be used with version numbers that aren't necessarily for the game, and so is its Lua counterpart that someone eventually added later (wesnoth.compare_versions). 20180523 17:56:46<+discordbot> The tl;dr is that the semantics for version number comparisons need to remain unchanged since that's actually an API. 20180523 17:57:32<+discordbot> The 0.0.1 < `0.0.1anything 20180523 17:57:43<+discordbot> The 0.0.1 < 0.0.1anything provision in particular. 20180523 17:59:19<+discordbot> hmm so x.y.z-something is considered AFTER x.y.z? 20180523 18:00:04<+discordbot> I don't know what else the less-than sign could possibly mean. 20180523 18:01:01<+discordbot> And it can't be changed to BEFORE for some kinds of "anything"? 20180523 18:01:51<+discordbot> That's an intrusive change that basically sends out the message that version_info understands natural language. 20180523 18:02:07<+discordbot> hmm not exactly 20180523 18:02:24<+discordbot> we can make this a convention so "-" denotes pre version 20180523 18:02:35<+discordbot> or only "-rc" and "-dev" 20180523 18:02:39<+discordbot> That's the kind of API change I just said is undesirable. 20180523 18:03:00<+discordbot> I already said the whole thing is exposed to add-on content. 20180523 18:03:29<+discordbot> so it can be used on addon content versions rather than wesnoth versions? 20180523 18:03:37<+discordbot> GAH. 20180523 18:03:45<+discordbot> I just explained that it was originally created for add-ons! 20180523 18:04:16<+discordbot> And that then it got used for the game as well! And that add-ons can use it to handle add-on versions as well as engine versions! 20180523 18:04:28<+discordbot> hmm ok 20180523 18:04:35<+discordbot> And that there are two explicit entry points to the API. 20180523 18:05:02<+discordbot> And a third implicit one, which is the version attribute for add-ons on campaignd. 20180523 18:32:42-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has quit [Remote host closed the connection] 20180523 18:32:50-!- janebot [~Gambot@unaffiliated/gambit/bot/gambot] has joined #wesnoth-dev 20180523 19:12:22<+discordbot> Memory leak occurs when building the teleport map. 20180523 19:12:41<+discordbot> (pathfind::get_teleport_locations()) 20180523 19:13:45<+discordbot> Can we please get some kind of fix or workaround for #2912? I don't like the tone of the latest person encountering it but it's not an acceptable bug to have in a stable release either. 20180523 19:24:42< irker286> wesnoth/wesnoth:master gfgtdf fb54b0376d don't autosave after using 'back to turn AppVeyor: All builds passed 20180523 19:43:34< gfgtdf> jykive: is tham memleak only on master ? 20180523 19:43:48<+discordbot> No. I'm testing in the 1.14 branch. 20180523 19:44:55< gfgtdf> i tried by creatign a siver mage and hovering over it a few times but i could not reproduce. 20180523 19:45:44<+discordbot> It occurs in the map submitted to Discord chat last weekend. 20180523 19:45:54<+discordbot> The map is very large and has a ton of towns. 20180523 19:48:41< gfgtdf> at wesnoth-dev ? 20180523 19:53:56<+discordbot> Vultraz uploaded a copy to this channel in Monday at 02:16 UTC. 20180523 19:57:34< gfgtdf> hmm ok 20180523 20:00:56< gfgtdf> anyways my first suspect would be the formula intepeter since it already had memleaks in the past. 20180523 20:01:33< gfgtdf> if you matthiaskrgr niceley he might also help you with his debugging tools. 20180523 20:01:56<+discordbot> I'm not sure if the tools would help. 20180523 20:02:17<+discordbot> It's possible that this isn't a true memory leak, but just some data structure that grows unbounded. 20180523 20:02:25<+discordbot> Automated tools don't detect those. 20180523 20:02:54< gfgtdf> ye thats true 20180523 20:19:22< zookeeper> totally random question without looking at the map: are a lot of the villages close to each other? i'd _think_ that the pathfinding options would grow massively if there's a lot of villages just a few hexes apart so that the silver mage could perform normal moves to get from one teleport location to another, etc. 20180523 20:20:31<+discordbot> Yes, it has clusters of villages very close to each other. 20180523 20:22:49< zookeeper> any idea if that could actually be a contributing factor? 20180523 20:23:16<+discordbot> Yes, I think it's a contributing factor. 20180523 20:23:16< zookeeper> it could also be irrelevant, i'd imagine. 20180523 20:23:30<+discordbot> Like I said, the mage in question can reach almost 6000 hexes in one turn. 20180523 20:25:56< gfgtdf> really? when i loaded the map in question the mage had onyl 2 moves left. which is enoght for move on village+teleport, unless it automaticially calclates multiturn moves, clusterd villages shouldnt matter then 20180523 20:26:06< gfgtdf> only* 20180523 20:26:22< gfgtdf> and i still got the lag 20180523 20:26:22<+discordbot> The map has a lot of Silver Mages. 20180523 20:26:34<+discordbot> I tested with one near the top-right corner of the map. 20180523 20:27:02< gfgtdf> i tsted with the mage that had an autovove saved (because the game crollscrolled to that position then) 20180523 20:48:24-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 20:48:53< gfgtdf> are we waiting for somethign particular before tagging 1.14.2? 20180523 20:52:50<+discordbot> Yes. 20180523 20:53:09<+discordbot> I don't remember what it is now and vultraz is not online at the moment and rarely bothers to read logs. 20180523 20:54:00<+discordbot> (Hi future vultraz if you're reading this, your face is dumb xoxo) 20180523 20:54:11<+discordbot> (Just trying to see if he actually doesn't read logs.) 20180523 20:55:06<+discordbot> (Now make the chat move along so my bait isn't on his screen by the time he wakes up.) 20180523 20:55:59<+discordbot> Does anyone other than him know if any changes to the add-ons management logic were made in 1.13.x that could result in this? https://forums.wesnoth.org/viewtopic.php?p=628693#p628693 20180523 20:56:00< gfgtdf> is it something he is workin on himself? 20180523 20:57:01<+discordbot> Also should I repeat myself re #2912? 20180523 20:57:29<+discordbot> It's not great that we have a mainline campaign that just crashes the moment you reach a cutscene if your window isn't large enough. 20180523 20:58:07<+discordbot> I know vultraz is just hoping that the underlying bug is fixed some day, but I believe we ought to do better for a stable release. 20180523 21:03:25< gfgtdf> i agree that it's an important issue, but its totally not the code i am curently working on, so not much i can do. 20180523 21:03:59<+discordbot> Anyone at all, perhaps zookeeper since he maintains TRoW I believe. 20180523 21:04:23<+discordbot> I think we'd rather have Rithrandil's portrait cropped at the top than crash the game. 20180523 21:06:03< zookeeper> well, we could just add a simple IPF chain to scale/crop it, as a temporary workaround. 20180523 21:07:37< zookeeper> but why are we having that problem again now, wasn't the same issue (huge portraits causing a crash/error) fixed a long time ago? 20180523 21:10:09< gfgtdf> the last comment here https://github.com/wesnoth/wesnoth/issues/1272#issuecomment-377991007 suggests ~SCALE_INTO 20180523 21:10:49< zookeeper> i'm right now checking what the max size can be, then i'll add some scale/crop thingy... 20180523 21:13:26< zookeeper> umm, i can't even reproduce. 20180523 21:14:29< zookeeper> also ffs could someone fix lua errors causing the rest of the event getting skipped already? 20180523 21:15:38< zookeeper> oh, hmm. it really requires a x720 resolution to trigger the bug, x600 works just fine. 20180523 21:15:50-!- Bhoren [~Bhoren_wh@2a01:e0a:c:2150:d19d:a3c:7917:5a2d] has quit [Quit: Leaving] 20180523 21:22:52<+discordbot> Probably because smaller resolutions trigger a different layout formula path. 20180523 21:23:38<+discordbot> (I know the [message] window doesn't use [resolution] but it does use formula logic to select the sizes of certain elements.) 20180523 21:24:40< zookeeper> mm-hmh. 20180523 21:25:10< zookeeper> so if i commit the IPF workaround now and head off to bed, will someone handle changelogging it? 20180523 21:28:57-!- travis-ci [~travis-ci@ec2-54-90-79-5.compute-1.amazonaws.com] has joined #wesnoth-dev 20180523 21:28:58< travis-ci> gfgtdf/wesnoth#1131 (1.14 - 18955d5 : gfgtdf): The build was broken. 20180523 21:28:59< travis-ci> Build details : https://travis-ci.org/gfgtdf/wesnoth/builds/382887651 20180523 21:28:59-!- travis-ci [~travis-ci@ec2-54-90-79-5.compute-1.amazonaws.com] has left #wesnoth-dev [] 20180523 21:29:35<+discordbot> Yeah, I can do that. 20180523 21:30:16< irker286> wesnoth: ln-zookeeper wesnoth:1.14 e41cd0b27f97 / data/campaigns/The_Rise_Of_Wesnoth/scenarios/ (05_The_Oldwood.cfg 07_Return_to_Oldwood.cfg): TRoW: IPF-downscale Rithrandil's portrait, as a workaround to #2912 https://github.com/wesnoth/wesnoth/commit/e41cd0b27f97ddaadf5bc471abf1a7a9e352debf 20180523 21:31:07< zookeeper> perhaps in theory one could check the current window size via some elaborate lua magic, but... 20180523 21:31:34< zookeeper> i guess that'll do for now, even if it causes a downscaling for the majority of users 20180523 21:32:37-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 21:35:00< zookeeper> well, i guess if _not_ using the minimal cutscene theme fixes the issue too, then that would be an alternative. 20180523 21:35:18<+discordbot> What's the portrait's actual height again? 20180523 21:35:24-!- travis-ci [~travis-ci@ec2-54-196-111-46.compute-1.amazonaws.com] has joined #wesnoth-dev 20180523 21:35:25< travis-ci> gfgtdf/wesnoth#1132 (1.14 - 30a21c5 : gfgtdf): The build was canceled. 20180523 21:35:25< travis-ci> Build details : https://travis-ci.org/gfgtdf/wesnoth/builds/382890940 20180523 21:35:25-!- travis-ci [~travis-ci@ec2-54-196-111-46.compute-1.amazonaws.com] has left #wesnoth-dev [] 20180523 21:35:57<+discordbot> Also, you can use [ci skip] in commit messages to skip building commits that don't introduce any changes that matter to the engine or to loading Wesnoth. 20180523 21:36:04<+discordbot> (On its own line somewhere.) 20180523 21:36:24-!- boucman [~rosen@wesnoth/developer/boucman] has quit [Remote host closed the connection] 20180523 21:38:10< zookeeper> 500x727, but yes, i was too hasty... i can keep it as high as 700 without problems 20180523 21:39:54<+discordbot> Hm. 20180523 21:40:03<+discordbot> That's certainly very tall. 20180523 21:40:40< irker286> wesnoth: ln-zookeeper wesnoth:1.14 4d92cc2395e4 / data/campaigns/The_Rise_Of_Wesnoth/scenarios/ (05_The_Oldwood.cfg 07_Return_to_Oldwood.cfg): TRoW: Increase Rithrandil's workaround portrait height to 700 https://github.com/wesnoth/wesnoth/commit/4d92cc2395e45e84f34b93b86015c2400cff40ce 20180523 21:41:39< zookeeper> yeah. i don't know why LB gave it such an unusual height anyway, seeing how it could have just been 700. 20180523 21:42:16< irker286> wesnoth: Iris Morelle wesnoth:1.14 1b04b9da4d04 / changelog.md: Changelog entry for the #2912 work around https://github.com/wesnoth/wesnoth/commit/1b04b9da4d049fddd6a23a1bd9f80fa02537ed5e 20180523 21:44:11-!- vn971 [~vasya@94.158.103.15] has joined #wesnoth-dev 20180523 21:56:30-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 245 seconds] 20180523 22:00:31-!- travis-ci [~travis-ci@ec2-54-242-49-100.compute-1.amazonaws.com] has joined #wesnoth-dev 20180523 22:00:32< travis-ci> gfgtdf/wesnoth#1134 (1.14 - 57a0297 : gfgtdf): The build was fixed. 20180523 22:00:32< travis-ci> Build details : https://travis-ci.org/gfgtdf/wesnoth/builds/382895242 20180523 22:00:32-!- travis-ci [~travis-ci@ec2-54-242-49-100.compute-1.amazonaws.com] has left #wesnoth-dev [] 20180523 22:19:13< irker286> wesnoth: gfgtdf wesnoth:1.14 ee50171d13c7 / src/ (carryover.cpp game_initialization/connect_engine.cpp): fix different carryover when loading startofscenario-save https://github.com/wesnoth/wesnoth/commit/ee50171d13c7f20a3c8147201c2b78ab54fafae3 20180523 22:51:45-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 22:52:52<+discordbot> What noone thought about so far.... the hotkey changes outdate the manual, and the manual can't be updated because we are in string freeze 20180523 22:58:21< Ravana_> I understand very few hotkeys were removed 20180523 23:12:11<+discordbot> Why on earth is the bottom row of the Preferences dialog layout so convoluted? 20180523 23:12:31<+discordbot> (Do. NOT. Touch it. I'm messing around with it right now.) 20180523 23:14:35<+discordbot> bottom row? 20180523 23:14:39<+discordbot> Bottom row. 20180523 23:14:43<+discordbot> The bottom-most row. 20180523 23:14:45<+discordbot> which panel? 20180523 23:14:57<+discordbot> None of them. I said the dialog's bottom row. 20180523 23:15:30<+discordbot> I... have no idea 20180523 23:15:33<+discordbot> there's no good reason 20180523 23:15:41<+discordbot> That's funny because you're the one who wrote it. : 20180523 23:15:58<+discordbot> Actually, you wrote it first 20180523 23:16:06<+discordbot> (Also for the people who were wondering if vultraz read logs, the answer turns out to be no.) 20180523 23:16:19<+discordbot> Yes, but you did most of it. 20180523 23:16:19<+discordbot> Nothing wrong with my face 20180523 23:16:26<+discordbot> Wow. 20180523 23:16:32-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 23:16:53<+discordbot> Then why when people tag you while you are asleep we tend to never receive an answer unless I specifically call attention to your mentions? 😛 20180523 23:17:27<+discordbot> I skim the backlog 20180523 23:17:33<+discordbot> Most of the time 20180523 23:17:43<+discordbot> If you really want to get my attention you must use @ 20180523 23:17:52<+discordbot> I said tag. 20180523 23:17:56<+discordbot> @Vultraz As in this. 20180523 23:18:32-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev 20180523 23:18:37<+discordbot> Would it be overkill to include the version number along with the version info button at the bottom left? 20180523 23:18:46<+discordbot> Of Preferences, that is. 20180523 23:19:05<+discordbot> yes 20180523 23:19:22<+discordbot> Okay, I'll only include the button then. 20180523 23:19:35-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Remote host closed the connection] 20180523 23:26:55< irker286> wesnoth: gfgtdf wesnoth:master fef4254c84c3 / src/ (carryover.cpp game_initialization/connect_engine.cpp): fix different carryover when loading startofscenario-save https://github.com/wesnoth/wesnoth/commit/fef4254c84c3e7c8a63250809408971e0c795b1e 20180523 23:26:57< irker286> wesnoth: Charles Dang wesnoth:master 9bf7a4c8fb4c / changelog.md data/gui/window/wml_message.cfg: GUI2/Message: tweaked Magic Numbers to fix layout failure with large portraits https://github.com/wesnoth/wesnoth/commit/9bf7a4c8fb4cf4c0ace3e745c855a26fcdf04626 20180523 23:27:04< irker286> wesnoth: Charles Dang wesnoth:1.14 489fbccba15f / / (3 files in 2 dirs): TRoW: revert workaround for #2912 https://github.com/wesnoth/wesnoth/commit/489fbccba15fafe2c1a4a3b22ec2c2007bddcdfe 20180523 23:27:07< irker286> wesnoth: Charles Dang wesnoth:1.14 d379abc4a95a / changelog.md data/gui/window/wml_message.cfg: GUI2/Message: tweaked Magic Numbers to fix layout failure with large portraits https://github.com/wesnoth/wesnoth/commit/d379abc4a95abd10831f29a43a2bd9cd97cebc2a 20180523 23:27:28<+discordbot> gfgtdf: Are you going to grace us with a changelog entry for that fix I spotted you pushing an hour ago, btw? 20180523 23:28:06<+discordbot> I tested my fix at all resolutions up to around what I have maximized 20180523 23:28:54<+discordbot> it's not a perfect solution but... 20180523 23:28:57<+discordbot> I wouldn't have bothered zookeeper if you had mentioned you had a fix incoming. 20180523 23:28:58<+discordbot> it works 20180523 23:29:24<+discordbot> I started working on a fix an hour ago after I saw the hack. 20180523 23:29:36<+discordbot> I mean, the bug was open before 1.14.0. 20180523 23:29:44<+discordbot> But okay. 20180523 23:30:33<+discordbot> it wasn't what I originally imagined the fix would be 20180523 23:33:12<+discordbot> 20180523 19:01:23 error config: malformed user credentials (did you manually edit the preferences file?) 20180523 23:33:24-!- heirecka [~heirecka@exherbo/developer/heirecka] has quit [Remote host closed the connection] 20180523 23:33:50<+discordbot> Do we still need that crummy data value type hack for usernames? 20180523 23:34:11<+discordbot> not that I know of 20180523 23:34:18<+discordbot> (I have a feeling everyone here is too new to know why I implemented it in the first place.) 20180523 23:34:34<+discordbot> I don't know why it was implemented in the first place 20180523 23:35:05<+discordbot> Usernames like "01234" or "0x4040" and others were being converted to their "natural" serialization as if they were integer values. 20180523 23:35:50<+discordbot> In this case that'd be "668" and "16448". 20180523 23:36:11<+discordbot> Same applied to passwords. 20180523 23:36:58<+discordbot> Usernames are still stored as WML but I don't know if the config implementation still uses atoi's semantics. 20180523 23:38:03<+discordbot> (Or strtod's, for that matter.) 20180523 23:38:14<+discordbot> We don’t use atoi anymore. 20180523 23:38:18<+discordbot> The semantics. 20180523 23:44:41< irker286> wesnoth: Iris Morelle wesnoth:master e8c803d4267a / / (3 files in 3 dirs): gui2/preferences: Add version info button https://github.com/wesnoth/wesnoth/commit/e8c803d4267aa0fb5d8693fb02dca7e6d2e7a0e8 20180523 23:44:46< irker286> wesnoth: Iris Morelle wesnoth:1.14 cfc0c4f8e1e9 / / (3 files in 3 dirs): gui2/preferences: Add version info button https://github.com/wesnoth/wesnoth/commit/cfc0c4f8e1e93d1adc82c1178f59579735a22a08 20180523 23:47:16< irker286> wesnoth: Iris Morelle wesnoth:1.14 d68815f53849 / players_changelog.md: Add a couple of entries to players_changelog https://github.com/wesnoth/wesnoth/commit/d68815f5384932a9afb0e187093f21c9118ec6a4 20180523 23:47:50< irker286> wesnoth: Iris Morelle wesnoth:master 2bc7be607e93 / players_changelog.md: Add a couple of entries to players_changelog https://github.com/wesnoth/wesnoth/commit/2bc7be607e938eba916cb3ef1dd4f2bf5fabfb06 20180523 23:48:41<+discordbot> you forgot the border for the info button 20180523 23:49:18<+discordbot> Oh. 20180523 23:50:42< irker286> wesnoth: Iris Morelle wesnoth:1.14 f314d48f1d9f / data/gui/window/preferences.cfg: gui2/preferences: Add margin around the version info button https://github.com/wesnoth/wesnoth/commit/f314d48f1d9fa84a3516204767479a92f682a21f 20180523 23:50:50< irker286> wesnoth: Iris Morelle wesnoth:master f30634842eb9 / data/gui/window/preferences.cfg: gui2/preferences: Add margin around the version info button https://github.com/wesnoth/wesnoth/commit/f30634842eb92ce31a5d35badec3ae97f4550eac 20180523 23:52:33-!- celticminstrel [~celmin@unaffiliated/celticminstrel] has joined #wesnoth-dev 20180523 23:57:03<+discordbot> Semi-periodic reminder that this is a thing: https://forums.wesnoth.org/viewtopic.php?p=628710#p628710 --- Log closed Thu May 24 00:00:01 2018