Time Nick Message 06:16 Zeno` hi hmmmm 06:16 hmmmm hello zeno` 06:18 Zeno` hmmmm see this: https://github.com/minetest/minetest/issues/1559 ? 06:19 hmmmm it's really not that simple 06:19 hmmmm that's caused by some very obscure bug 06:19 Zeno` I think I know one reason that causes that behaviour: changing the mapgen params on an already generated map (I mean a map that already has areas mapgenned) 06:19 Zeno` I can duplicate it 06:19 hmmmm it's not like we say, "oh! let's keep running the cave generation in old chunks!" 06:20 Zeno` but, that's just *one* potential reason 06:20 Zeno` Just thought I'd let someone know 06:20 hmmmm if you change the mapgen params on a map that's already been created, it'll ignore those and keep using the parameters it had been created with 06:20 Zeno` it doesn't seem to for me 06:22 Zeno` I change the values in world.mt to crazy values, I get crazy landscapes 06:22 Zeno` *shrug* 06:22 Zeno` I'll investigate more later 06:23 Zeno` are the mapgen parameters duplicated in the database (leveldb) ? 06:29 hmmmm no 06:29 hmmmm the database file only contains map block data 06:29 hmmmm the map metadata is stored in map_meta.txt 06:42 Zeno` so how does changing it "it'll ignore those and keep using the parameters it had been created with" 06:46 hmmmm you might be confused 06:47 hmmmm you're supposed to change mapgen parameters in the minetest configuration file 06:47 hmmmm mapgen parameters that are stored in map_meta.txt are not meant to be modified by end users and are there to preserve map settings 06:48 Zeno` I bet many people do change them, supposed to or not 06:48 Zeno` I'm just saying, that's all. :) 06:49 hmmmm so you want them to be hidden better? 06:49 Zeno` I don't think that's a solution any more than you do :p 06:49 Zeno` I merely raised the point as a "potential" source for these hard to find / obscure bugs 06:49 Zeno` which may not be "bugs" as such 06:51 Zeno` I.e. I'm just suggesting that maybe (at least once in a while) such things may be end user error 06:51 hmmmm most "bugs" end up as user error because too much trust is placed in the user 06:51 Zeno` Well, it's open source. You can't hide things 06:51 hmmmm it just doesn't seem right to restrict those who know what they're doing thanks to people who don't 06:52 Zeno` I'm not suggesting that as a solution :) 06:52 hmmmm major web browsers are removing features 06:52 Zeno` I'm suggesting that maybe there is no bug at all 06:52 hmmmm that's one way 06:52 Zeno` Yeah, a very annoying way 06:52 hmmmm minetest is nowhere nearly as mature as, say, firefox though 06:52 hmmmm removing features is possibly the worst thing that could be done 06:53 Zeno` I agree 06:53 twoelk if it is a user mistake it can be described in some doc, like see we told you not to do that 06:53 Zeno` twoelk, so perhaps the solution is just that... 06:53 Zeno` *if* that is the source of the error 06:54 twoelk but from a user about a problem related or not, I played a lot with this file described here https://forum.minetest.net/viewtopic.php?p=150822#p150822 06:54 Zeno` I know that it *can* cause that bug... I don't know if that's what people are seeing it are actually doing that though 06:54 twoelk I moved around a very high building that triggered some mapgen a lot of times 06:55 twoelk sometimes caves but often only grass toppings 06:55 twoelk I tried preloading the area, flying around whatever 06:56 twoelk It seemed that mapgen got interested again once the area contained some stone 06:57 hmmmm that can happen if a block is already exists, is generated, and stored in the database, but is not marked with the "generated" flag 06:57 hmmmm which is caused by some very difficult to track bug 06:58 Zeno` potentially map_meta.txt has another subtle error though... the precision of floats that the Lua lib is compiled with (you can, for example, change it to compile with float instead of double [or the other way around... I can't remember the default]). This would *potentially* cause very small errors or differences in real numbers 06:58 hmmmm that's pretty moot because floating point numbers are expected to have errors in them 06:59 twoelk so it might be something saved wrong to the database? 06:59 twoelk has this been reported from all database kinds then? 06:59 hmmmm this could happen if the block is unloaded as it's being generated 07:00 Zeno` hmmmm, that's correct but if you round a float vs. a double (both initialised with a constant) there *can* be differences in the result (easily) 07:00 hmmmm it has nothing to do with the database, but rather an internal logic error or maybe a race condition 07:01 Zeno` I'm no so sure about the race condition myself. The simplest answer, to me, is that people are changing map_meta.txt by hand (I've done it) 07:01 Zeno` There's nowhere saying it's a bad idea, so why not? 07:02 twoelk I definitly did not change any map_meta.txt on that map 07:03 twoelk If mapgen is not to touch allready generated area, I would rather think it is bypassing that 07:04 twoelk maybe as a user comes along it imposes a change by the user 07:05 Zeno` I wouldn't *expect* changes to map_meta.txt to affect already generated areas either 07:05 hmmmm somebody brought up mapgen params 07:05 hmmmm i was just explaining that 07:05 Zeno` I can't think of anything I'd *expect* to change already generated areas 07:05 hmmmm changes to map_meta.txt actually might do this however. in particular, chunksize 07:06 Zeno` hmm, yes 07:06 Zeno` or seed 07:06 hmmmm no, not seed 07:06 Zeno` well, not seed if chunsize remains constant 07:12 twoelk regarding the tall building, when it happened, caves would munch any node kind but only "natural" nodes such as stone or sand would be replaced, while the grass topping would be placed ontop of anything like them torches at the top 07:13 hmmmm it could be that the block is silently failing to save 07:13 twoelk so the glass parts might be gone but not replaced 07:13 hmmmm if the server in question is out of date 07:14 twoelk so it should happen more often on old servers/hardware? 07:15 hmmmm old servers that haven't been updated to prevent this kind of thing 07:15 hmmmm actually come to think of it, this is exactly what would be caused by a block that failed to save before the semi-recent fix 07:16 twoelk well my playing with the tall building was a local singleplayer map made that day although on an old computer ;-P 07:19 twoelk so a map should, for example be converted to another database and back to force a rewrite that might fix old issues 07:21 twoelk I don't think this will cover all cases, something is happening on new maps as well 07:29 Zeno` ugh 07:30 Zeno` so bug fixes are now "enhancements" hehe 07:43 * twoelk wonders what events actually trigger mapgen 07:55 Zeno` class EmergeManager looks after it all (I think) 07:56 Zeno` and MapgenFactory 07:56 Zeno` lol @ "factory" 07:56 Zeno` Java style 11:09 Exio4 OOP design pattern's book was written in C++ if i remember correctly 11:12 sapier "design patterns" ... well if ppl know what those are good for they may help discussing about things ... but at least what I experienced lot of ppl not knowing what they do hide their missing knowledge behind shiny pattern names 11:16 Zeno` meh, sedgewick? 11:17 Zeno` sapier, why is my buf fix now an... enhancement? 11:17 Zeno` bug* 11:18 sapier which one? 11:18 Zeno` can't remember... l-system trees not being random 11:19 sapier adding randomness seemd to be an enhancement to me as it does work without too ... yet the result is way better with it ... isn't it? 11:19 Zeno` whether it's a bug or not is probably debatable, but the biggest mod that uses l-system tree is moretrees 11:19 Zeno` no, without that change all the trees look the same because the seed is the same 11:20 sapier that's not nice but still correct 11:20 Zeno` unless VE changes moretrees to pass a random see via Lua 11:20 Zeno` if it's not a bug then I think the documentation should be updated 11:21 sapier hmm are you trying to tell me it's a core workaround for a mod bug? 11:21 Zeno` to stress that unless random_seed (or whatever it's called) is supplied via Lua all the trees will look the same 11:21 sapier can random seed be passed from lua there? 11:21 Zeno` yeah it can 11:22 sapier if I pass a random seed will this one be used after your fix? 11:22 sfan5 sapier: if you want to save space for minetest then make the button have less width to not make them look so compressed 11:22 Zeno` yes 11:22 Zeno` if you explicitly set random_seed then that will be used 11:22 Zeno` if you do not set a seed then the seed with will random 11:22 sapier Zeno`: ok then your fix at least doesn't break things 11:23 sapier because e.g. if someone intentionally wants to create exactly same bugs a forced random seed would be bad 11:23 sapier sfan5: where did I want to save space? 11:23 Zeno` I made sure of that :) I debated for many days whether or not this was needed. A fixed seed is useful for debugging 11:24 sfan5 [...] 4) intentional to save room for minetest 11:24 Zeno` But you can intentionally create the same PNRG sequence after the patch (by supplying a fixed seed) 11:24 sfan5 jump and descend buttons 11:25 sapier sfan5: ahh now I'm back with you ... problem with "less width" is my fingers have some fixed width while I can place them at the lower end of screen I can't cut them into pieces ... well maybe I could but I wont do for sure ;-P 11:25 Zeno` At the moment every mod I have looked at (mostly moretrees I admit) assumes that the trees will be random in some way; they won't currently because the PNRG seed is always 0 (because none of the mods I've looked at supply a random seed from Lua) 11:26 Zeno` why do I keep typing PNRG? lol 11:27 sfan5 Zeno`: it genrates Pseudo-numbers! :P 11:27 Zeno` yeah I should be type PRNG though :p 11:27 sapier sfan5: I understand your issues but changing that gui should only be done by checking more then one device, I understand that this ui isn't perfect maybe not even good for your device. Yet I refuse to completely optimize it to your device if this means breaking other devices. 11:28 Zeno` in any case, either the patch needs to be applied or Lua API updated. Either will work. 11:28 Zeno` Lua API docs* 11:28 sfan5 sapier: just do whatever you want to make the button not look compressed 11:28 sfan5 ugly buttons is far worse than buttons that are too big or too small 11:29 sapier if buttons cover 2/3 of screen that's bad too 11:29 Zeno` If the patch is not applied VE will just have to update moretrees (although I think I agree with her that setting the seed in C++ is probably more efficient) 11:29 sfan5 thats not what I mean 11:29 sapier especially if it's a tablet screen 11:29 sfan5 buttons being _a bit_ too small or too big is not as bad as buttons looking ugly 11:29 sapier sfan5: but that's why this button placement was initially done 11:30 sfan5 whatever 11:30 sfan5 just try to make them look less ugly 11:30 sapier sfan5: If I could do that I would ;-) but I'm still no designer and wont get one anytime soon ;-) 11:30 Zeno` I have a saying: "ugly is the eye of the beholder" 11:30 Zeno` I came up with that myself 11:30 sfan5 button.width = button.width * 0.75 11:30 Zeno` no help 11:30 sfan5 that should solve it 11:31 sapier no 11:31 sapier that'd waste room on screen 11:31 sapier not fix the issue 11:31 sapier your issue is "buttons are ugly" but your solution is "change the ui" ... that's crap 11:32 sfan5 how would you fix that w/o changing the UI? 11:32 sapier a different pictogram maybe? 11:32 sapier something not looking compressed but showing the meaning 11:34 sapier btw someone wanted to implement some sort of "drag dot" ui ... I wouldn't have problem with replacing the ui in total ... I just don't have time to do it myself ... and even less interest in fighting for that ui ... because as usual in ui there'll be as many ppl not liking as liking it 11:35 Zeno` why should there be fighting? :( 11:35 sapier this is minetest 11:42 Zeno` oh yeah 11:42 Zeno` forgot 11:50 sapier but still I know android ui isn't perfect I have no problem replacing the graphics with better ones but for ui I changes I demand more then one opinion about one particular device 11:56 sapier hmm maybe we should create a gallery how things look like on different devices, as screenshots aren't enough to show 11:56 sapier we'd need to scale the image to devices screen dpi 11:56 Zefram_Fysh that sounds useful 11:56 sapier well but a lot of work too 11:57 Zefram_Fysh it's annoying how those two things are often correlated 11:57 sapier ui highly depends on screen size and dpi and we don't have any experience in those things yet 11:58 sapier minetest just didn't handle them a few versions ago and the current way of doing it is obviously bugged as hell 11:59 sapier I wonder if it's worth adding scalable non freetype fonts 11:59 Zefram_Fysh would making freetype a hard dependency be a problem? 12:00 sapier well depends 12:00 sapier freetype implementation for irrlicht is bugged as hell 12:01 Zefram_Fysh hmm 12:01 sapier it doesn't handle a lot of font errors 12:01 sapier except of crashing of course 12:01 Zefram_Fysh maybe it would be worth fixing that 12:01 sapier hmm 12:02 sapier well looking at our fallback freetype font 4.4 mb ... adding 3mb of scalable non freetype fonts doesn't look as bad as I thought 12:03 Zefram_Fysh sure, the data size isn't terrible, but it's a pain to maintain parallel implementations of the font stuff 12:03 Zeno` maybe *that* doesn't look bad, but the fonts will 12:03 Zefram_Fysh fewer options to consider for the gallery of UI appearance 12:04 sapier as irrlicht doesn't support freetype anyway we'd always have a fallback 12:05 Zeno` freetype rendering is probably more complicated than the entire irrlicht code base 12:05 Zeno` there is some serious crap that needs to be done to render fonts correctly 12:06 sapier possible but I won't rewrite freetype for sure nor will I look for errors in there ;-) 12:06 Zeno` the freetype mailing list is very active. If an error is found it's usually fixed within days 12:07 sapier freetype isn't the problem irrlicht is 12:08 Zeno` In that case my answer to "sapier> I wonder if it's worth adding scalable non freetype fonts" would be a definite NO 12:08 Zeno` they'll almost certainly look like crap 12:09 sapier for what reason? right now non scalable fonts are defacto same as freetype fonts 12:09 sapier with one small difference 12:09 sapier they're prerendered while freetype fonts are rendered otf prior beeing passed to irrlicht 12:10 sapier and no they don't look like crap they look almost identical 12:11 sapier the only difference is freetype will render any font sizes .. while non ft fonts will have to be there in some chosen intervals ... most likely the ones the irrlich font tool suggests 12:12 sapier but without scaling formspecs discussion is almost useless 12:12 sapier only android port would benefit 12:13 Zeno` If I build with freetype disabled though the fonts are terrible 12:13 Zeno` why is that? Hinting or something going wrong? 12:14 Zeno` They don't look the same at all to me 12:15 sapier because the font used for non freetype mode is crap ;-) 12:15 Zeno` heh, maybe 12:17 sapier https://github.com/minetest/minetest/pull/1561 you can try this one, to see fonts scaling set screen_dpi to a higher value. Yes menu will be broken I'm adressing this in another pull ... even more controversial as it fixes formspec positioning mechanism too 14:44 sapier https://github.com/minetest/minetest/pull/1565/files realbadangel_ please base your subtab thingy for settings on this one 16:21 sfan5 this idea just came up in #minetest: http://sprunge.us/XMfU 16:22 sfan5 and having opinions would be quite nice 16:26 PilzAdam why abuse the serverlist for that? 16:28 sfan5 how else do you tell people with old clients to update? 16:30 PilzAdam a message box at startup would be better 16:30 sfan5 how exactly to you think about implementing that into versions already released? 16:31 sfan5 s/to/do/ 16:31 Calinou more important would be server banning from master 16:31 Calinou but you can add update notification too :) 16:31 Calinou Debian will have to censor it out 16:31 Calinou (idea: add minetest.conf setting for update check, to make it easy to remove by Debian and such) 16:33 PilzAdam sfan5, it would obviously be for future outdated versions 16:34 VanessaE PilzAdam: so, in other words...useless for the current situation. 16:34 PilzAdam also there is actually nothing wrong with using 0.4.8, so "blocking" the serverlist with entries isn't good 16:34 VanessaE and there is. 16:34 VanessaE old clients can't use some modern features. 16:35 VanessaE hell even your/my signs don't work on certain old clients 16:35 PilzAdam thats only if old clients connect to new servers 16:35 PilzAdam they should get a message from the server then 16:35 VanessaE yeah, and? 16:36 VanessaE how are you gonna get a message to the old client if it doesn't support any message-passing protocol? 16:36 PilzAdam minetest.chat_send_player()? 16:36 VanessaE server-side mods aren't allowed to detect client versions 16:37 VanessaE unless this was changed recently, which I doubt. 16:37 PilzAdam also doesn't the client get a warning anyway if connecting to a server with a different protocol version? 16:38 VanessaE only if the client isn't somehow patched to work around that. 16:38 VanessaE we're not talking protocol differences, we're talking about clients doing shit just plain wrongly that is otherwise theoretically "right" 16:38 VanessaE i.e. you send a block to the client, it displays it upside down or something 16:39 VanessaE s/block/block's textures/ 16:40 PilzAdam add a sign "If you see this block upside down upgrade your Minetest version" to the spawn in your server ;-) 16:40 VanessaE hard to make that work if the user can't read the sign :) 16:41 VanessaE (that's a real example of such a failure - because the texture for the text is blacked-out due to some GPU/driver bug in iOS devices) 16:42 PilzAdam btw, why are blocks displayed upside-down when old clients connect? 16:42 PilzAdam what broke it and why wasn't a compatibility layer added? 16:47 VanessaE the upside down block was hypothetical 16:47 VanessaE the sign one was real 16:48 VanessaE the problem is these "old clients" are often tablets/phones running not Minetest but buildcrap 16:48 VanessaE we can't ban them 16:48 VanessaE we can't patch them 16:48 VanessaE we can't kick them off 16:48 VanessaE and we can't send server-->client messages reliably 16:49 VanessaE but we CAN use the server list to inform users. 16:49 VanessaE and there's little or nothing that the buildcrap guys can do to "protect" against that without totally breaking the server list. 16:51 PilzAdam ah, so you want to do what the main reason is why mods are not allowed to get the client version 16:51 VanessaE uh, no 16:52 VanessaE the main reason for mods not being given server/client version is, as I thought, to prevent general breakage or reliance on versioning. 16:52 VanessaE this is an upgrade notice 16:52 VanessaE big difference 16:53 sfan5 more important would be server banning from master 16:53 sfan5 we can already do tha 16:53 sfan5 t 16:58 Calinou ah 17:29 sapier imho no matter what we do (maybe except of autoupdating user clients to master) wouldn't help VanessaE as she can't stick to a stable version but always has to use bleeding edge master code ;-P 17:30 sapier And for what I believe even if we did have a "stable" branch only merging bugfixes but no new features she wouldn't use it ... Would you VanessaE? 17:31 VanessaE you never know :P 17:32 Zefram_Fysh sapier: have you noticed a problem with Irrlicht tab controls, where the hotspots of the tabs don't match their visual extent? Irrlicht computes hotspots based on how the tabs would render using the font that's current when button is pressed, which doesn't necessarily match the font that was used for drawing 17:33 sapier Zefram_Fysh: that could be a problem 17:33 Zefram_Fysh I think it would arise with your formspec patch. it arises with my version, and it's just about the last thing standing in the way of me declaring my patch complete 17:34 sfan5 sapier: telling users to update doesn't help? 17:34 sapier well it doesn't happen for main menu right now as there the default font is used 17:34 sapier but it could happen for tabs in game 17:34 VanessaE sapier: the whole damn point is telling users to update if they're using a stable that's older than the current stable, THAT IS ALL. 17:34 Zefram_Fysh the Irrlicht innards are godawful wrt font selection. different GUI element types have different rules for it 17:34 VanessaE jesus christ, why is this so hard to grasp? 17:34 sfan5 lol 17:34 VanessaE and any proposal that involves changing the client WILL NOT WORK 17:35 Zefram_Fysh for me it's happening on the main menu, because that's not using the default font 17:35 sapier sfan5: I was talking about helping VanessaE ... to get this done we could write the text "update to latest master" hardcoded to menu ;-) 17:35 VanessaE sapier: s/latest master/0.4.10-release/ 17:36 sapier only until next nice feature in master making stable slightly less advanced then you'll complain about "those old outdated stable clients" 17:36 VanessaE sapier: I may code for dev, but my servers try to stick to working with *stable* 17:36 VanessaE but telling users it's olay to keep using 0.4.7? FUCK NO 17:37 sfan5 sapier: all I propose is displaying a message to people with <= 0.4.8 17:37 sapier Zefram_Fysh: for what I see the only option is make mainmenu use the default font ;-) 17:37 sapier sorry sfan5 I'm with pilzadam why abusing serverlist for things like that 17:38 sfan5 tell me how to do it otherwise 17:38 VanessaE sapier: in other words I don't give two shits if the CLIENT can run my mods, I only care if my servers can run them, and if the client can properly display what the server sends. that means using a -release client. and 0.4.10-release is more than new enough. 17:38 sapier Why do it at all? 17:38 Zefram_Fysh I guess I can contrive that, or I was also thinking about ways to make tab controls always use the default font regardless of context 17:38 VanessaE sapier: but anything older than 0.4.9 doesn't even work with those signs, and you KNOW that! 17:38 sfan5 sapier: because we can't be backwards compatible to every release 17:38 VanessaE you know how much of a pain in the ass it is to deal with old clients 17:38 sfan5 and the whole texture modifier stuff 17:38 VanessaE how they drag server performance down 17:38 VanessaE etc etc etf 17:38 VanessaE etc* 17:38 sapier Zefram_Fysh: that's crap you can't use small default font if all other menu fonts are scaled up 17:38 Zefram_Fysh it's also possible to leave the form's font selected in the skin, but that affects things like chat that use it 17:39 sapier that'd be one of those inconsistency things formspec already suffer from 17:39 Zefram_Fysh to get consistently not-crap results we're going to have to fix Irrlicht, I think 17:40 sapier so because you guys don't like 0.4.8 clients on YOUR servers we shall bug all 0.4.8 users which may even want to play on other servers running 0.4.8? 17:40 sapier Zefram_Fysh: you're right ... go back in time and fix irrlicht 1.7 17:40 sfan5 if they want to play on 0.4.8 servers they use 0.4.8 for those and 0.4.10 for the othersw 17:40 sfan5 -w 17:41 sapier sfan5: and how do you think they'll find 0.4.8 servers if you break serverlist? 17:41 sfan5 did you even read 17:41 sfan5 I proposed adding a notice at the top 17:41 sfan5 not replacing the servers 17:41 sapier if you want to tell ppl about not joining add it in description it's shown on client tab 17:42 sfan5 you don't want to understand it, do you? 17:42 sapier this description is server specific 17:42 sapier why bother all users because some servers want to tell them not to join? 17:43 sfan5 some servers? 17:43 sapier Zefram_Fysh: I was serious, you're right that this is a irrlicht bug but that doesn't help us as we won't tell ppl sorry but you need next version of irrlicht for about the next 2 years ;-) 17:44 sfan5 sapier: I heard you volunteer to maintain compatibility code (in mods) for 0.4.1? 17:44 sapier compatibility is mod owner resposibility ... noone forces you to do it 17:44 Zefram_Fysh yeah. I'm still looking at workaround options. it may be possible to wrap the tab control class or something 17:44 sfan5 are you serious? 17:44 VanessaE MOD OWNER? 17:45 VanessaE wtf? 17:45 sapier owner/writer/developer 17:45 sfan5 do you fucking expect any modder to support old crap 17:45 VanessaE now I know you're deliberately misunderstanding. 17:45 sfan5 do you? 17:45 VanessaE why should continue to cater to 0.4.4? 17:45 VanessaE or even 0.4.9? 17:45 sapier No I don't it's mod writers decision if she/he wants to be compatible or not this ain't a core matter 17:46 sfan5 so everyone should just set a higher minimum protocol version? 17:46 sapier wtf 17:46 VanessaE server list reads: Please update, your / client is outdated / 0.4.10 is available from / www.minetest.net / / / / etc etc etc 17:46 VanessaE that's ALL 17:46 VanessaE jesus fuck 17:47 sapier Just open up a message box on join "Minimum required client version is ..." 17:47 VanessaE sapier: that means changing the client. 17:47 VanessaE you can't reliably detect client versions from a mod. 17:47 sapier no it doesn't 17:47 sapier you don't 17:47 VanessaE then how do you propose to detect if the client is outdated? 17:48 Zefram_Fysh think sapier is proposing to show the message box to all clients, requiring the actual player to do the version comparison 17:48 sapier you don't have to, just post that message on initial connect, save whom has been told about and clear that list on each new version of your server 17:48 proller also lets start show ads in ingame serverlist 17:48 VanessaE that's STUPID 17:48 VanessaE shut up, proller 17:48 sfan5 what 17:48 proller sell viagra or drugs 17:48 sapier I'm not proud to tell it but proller is right this time 17:48 sfan5 do you expect people to know which version they use? 17:48 VanessaE this is not a slippery slope argument here 17:49 sfan5 what about the buildcrap people? 17:49 sfan5 and do you expect people to care if it "just works" 17:49 sapier I don't care about the buildcrap ppl as they're not using minetest, I can't patch their bugs away anyway 17:49 sfan5 no 17:49 VanessaE sapier: you can't tell every user to update. that's just plain stupid 17:49 sfan5 they're not using minetest 17:49 sfan5 ofc not 17:49 VanessaE and you can't patch existing clients 17:49 VanessaE that's ludicrous and impossible 17:50 VanessaE you have to do this either from the minetest server, which is impossible because we can't detect client versions, or from the master server list. 17:50 VanessaE s/we/mods/ 17:50 sapier VanessaE: nothing is impossible I gues we could even patch current clients if we abused some security flaws in minetest 17:50 VanessaE what? 17:50 sfan5 what 17:51 VanessaE YOU, talk about abusing a security flaw?! 17:51 VanessaE are you insane? 17:51 sfan5 ^ 17:51 sapier (19:49:53) VanessaE: and you can't patch existing clients 17:51 sfan5 and how do you reliably want to patch hundreds of different compiled versions by abusing e.g. a buffer overflow? 17:51 sfan5 pls tell me 17:51 sapier that's wrong ... we can but we wont do it that way 17:51 sfan5 (was probably mostly a joke anyway) 17:52 sfan5 and which security flaw are you talking about? 17:52 VanessaE you can't patch that which you don't have suitable code to patch 17:52 VanessaE you'd be better off inserting a patch guaranteed to crash the client with a "you're outdated, go fuck off" error. 17:52 sapier most easy way to patch minetest most likely would be make server send a model with prefixed minetst binary path 17:53 sapier hmm no might not work 17:53 VanessaE nice.. go ahead and put that in the public log. 17:53 sfan5 and how would that patch things? 17:53 sapier it'll be a little more complicated 17:53 sfan5 that isn't even a secrity bug 17:53 sfan5 irrlicht will say: nope, not a .x or .b3d, can't load that 17:53 sfan5 security* 17:53 proller sapier, start code binary update system now!! 17:53 sapier overwriting minetest.exe by a model sent from server wouldn't be a security bug? ;-) 17:53 sfan5 yes 17:53 sfan5 because everyone uses windows 17:54 sfan5 and what? 17:54 sfan5 how do you plan to write an arbritary file? 17:54 proller now biggest issue of minetest is lack of binary updates!! 17:54 sapier lol sfan5 irrlicht's file checks are a bad joke I'm quite sure I could make irrlicht execute almost any code by giving it a broken file 17:55 sfan5 they go to client/cache/media/ not client/cache/media/ 17:55 sfan5 sapier: yes, because ASLR and DEP are totally trivial to defeat 17:56 sapier to be honest for someone really knowing what to do they are 17:56 sfan5 *sigh* 17:56 sapier or having the right toolkit 17:56 sfan5 a toolkit doesn't help defeat ASLR 17:56 sapier yet back to our original topic abusing server list to provide information required by (some) server owners 17:56 sfan5 hmm 17:57 sapier No ... you didn't give me any reason to change my mind about that one 17:57 sfan5 maybe we should support 0.3 clients to 0.4 servers? 17:57 sfan5 that matches your principle of total backwards-compat. even if that is complete shit 17:57 sapier nope because 0.3 is not 0.4 and 0.5 will not be 0.4 too 17:57 * VanessaE triple-facepalms 17:58 sapier by now we handle things this way ... if this is right or not is not subject of this debate 17:58 PenguinDad sapier: and 0.4.7 is not 0.4.10 17:58 VanessaE sapier: s/some server owners/all server owners/ 17:58 sapier PenguinDad: true but by now compatibility break is supposed to cause 0.X.0 version number to change 17:58 VanessaE ALL server owners will have this problem eventually. 17:59 sfan5 <sapier> PenguinDad: true but by now compatibility break is supposed to cause 0.X.0 version number to change 17:59 sfan5 [citation needed] 17:59 VanessaE and frankly, we're getting tired of having to tell our users constantly to upgrade their busted, outdated, obsolete clients. 17:59 sapier see our issue list there are various things marked as 0.5 because of compatibility break 17:59 sfan5 hm 17:59 sfan5 we shouldn't add any features 18:00 sfan5 they break compat. too 18:00 sapier btw each new version has new nice features that can't be used with older clients we could just add a hardcoded text "update to latest client" 18:00 sfan5 hardcoded text where? 18:01 VanessaE sapier: upgrade to LATEST RELEASE 18:01 sapier right on top of mainmenu maybe 18:01 VanessaE jesus christ in a handbasket 18:01 sfan5 and how would that help people with 0.4.8? 18:01 VanessaE sapier: and how would existing clients see that? 18:01 sapier why do you wanna make 0.4.8 ppl to update? 18:01 VanessaE it HAS to be done with the minetest server or the master server list in order to reach all old clients 18:01 VanessaE because old clients are busted with some of the most basic crap 18:01 VanessaE and you can't fix them 18:01 sfan5 sapier: because minetest is not 100% backwards-compatible?? 18:02 sapier minetest isbackward compatible ... your mods aren't 18:02 sfan5 no 18:02 sfan5 minetest is not backwards-compatible 18:02 sfan5 every single addition to formspecs or texture modifiers are not backwards-compat. 18:02 sapier I can connect to latest server with 0.4.1 client and do all the basic stuff 18:02 sfan5 you can't blame mods to use them 18:02 sfan5 basic stuff? 18:02 sfan5 lol 18:02 proller basic lag 18:03 sfan5 you say we should reduce minetests function to basic stuff? 18:03 sapier if you don't like it this way we can just make server check client version and kick all old clients as this will be the only way to enforce compatibility 18:03 sfan5 wat 18:04 sfan5 WAT 18:04 VanessaE sapier: I repeat, I code for latest dev ON THE SERVER, but I try to make sure my code works for LATEST RELEASE ON THE CLIENT 18:04 sfan5 I don't want to enforce compat. 18:04 VanessaE what part of what don't you get? 18:04 sapier rigth now we did always try to add things in a way old clients still can do what they could do before 18:04 VanessaE in other words I don't give two shits if the client is newer than latest release usually. 18:04 VanessaE I only care if the server is. 18:04 sfan5 sapier: then you need to remove all of your formspec additionas 18:04 VanessaE only if there's some major feature that breaks things, as you well know. 18:04 sfan5 additions* 18:05 sapier sfan5: noone forces a server owner to use them? running a 0.4.10 server with old mods will result in 0.4.1 client to have ALL the features provided by server 18:05 sfan5 aswell as broken formspecs 18:06 sapier for sure not as those old mods wont send the new style formspecs 18:06 VanessaE sapier: so...what you're saying is server owners shouldn't bother with new features at all? 18:06 sfan5 please tell me someone who uses 0.4.1 mods on a 0.4.10 server? 18:06 sfan5 s/?// 18:07 sfan5 " have a choice" is the worst excuse 18:07 sapier sfan5: I don't care what you believe to be "most common usecase" ... for me it's enough someone could do it that way 18:07 sfan5 WAT 18:08 VanessaE sapier: look at the master list. 18:08 VanessaE now. 18:08 sapier and using those old mods is supposed to work perfectly fine 18:08 VanessaE right now. this instant. 18:08 sfan5 "someone could possibly use 0.4.1 mods with 0.4.10 so we don't need to tell people to update! lololololol" 18:08 VanessaE what is the oldest server version you see there? 18:08 VanessaE and what is the most common version? 18:08 sapier what exactly is this supposed to tell me except of some temporary snapshot? 18:08 VanessaE answer: 0.4.9, and 0.4.10, respectively. 18:08 VanessaE it's called statistics. 18:09 sfan5 "temporary snapshot" 18:09 sfan5 what 18:09 VanessaE no matter how often you look at this page, you'll get similar results. 18:09 sfan5 it's not like half of the servers might migrate to 0.4.1 mods next day 18:09 sapier so redcrab did update to 0.4.10 now? 18:09 VanessaE is redcrab on this list? no. 18:09 VanessaE we don't care what he runs 18:09 sfan5 oh 18:09 VanessaE he can't even bother to reply to his emails/forum posts anymore 18:09 sfan5 another great thing 18:09 sapier ok if you don't care why should I care 18:09 VanessaE because G*d damn it he runs 0.4.4 18:09 sfan5 "someone might be running 0.4.1 in his private network composed of 1 person, we need to support it" 18:10 VanessaE do you even realize how outdated that is? 18:10 VanessaE that is to minetest what kernel 2.2 is to Linux 18:10 sapier so why do we even discuss about compatibility if you drop it as soon as it fits your needs? 18:11 VanessaE because there has to be a threshold beyond which we don't care? 18:11 sfan5 Minetest has basic backwards-compat. but that doesn't mean we shouldn't tell people not to update to the latest version if they use an outdated one 18:11 VanessaE do you care if MOBF still works with 0.4.1? 18:11 sfan5 ^ 18:11 sapier then make us show this in version numbers and we're done 18:11 sfan5 I still need citation that the x in 0.x.y marks compatibility 18:12 VanessaE sfan5: only based on Minetest history, I think. 18:12 sapier VanessaE: yes I do provide old version of mob for those using old servers 18:12 VanessaE 0.3.x isn't compatible with 0.4.x, and 0.5.x is expected to break a lot of stuff as well 18:12 sfan5 sapier: you said something different 18:12 VanessaE sapier: no, I mean the current code. 18:12 VanessaE what's at git HEAD of your repo 18:12 sfan5 sapier: you said you expect basic stuff to work with 0.4.1 cleient <-> 0.4.10 servers 18:12 sfan5 then that needs to apply to mobf too 18:12 sapier no because I tell them what minimum version to use 18:12 VanessaE there. 18:13 VanessaE you said it. 18:13 sapier but I don't force anyone 18:13 VanessaE "minimum verson" 18:13 sfan5 wat 18:13 sapier and I don't add something to mobf preventing usage 18:13 VanessaE "minimum version" ** == "I don't care if this works with 18:13 sfan5 ^ 18:13 VanessaE and we're not proposing doing this with the server list either. 18:13 VanessaE there would be messages there, but the list will still work 18:14 sapier you're mixing up things you want us to force add something to serverlist which is not related to any version at all 18:14 VanessaE all of the servers will still be in the list! 18:14 sfan5 what 18:14 VanessaE there will just be a header above them that the user can see 18:14 VanessaE that's all 18:14 sfan5 not related to any version? 18:14 VanessaE a fake server or three in the list that is enough to display a message 18:14 sapier no matter what client someone will use no matter what server he's gonna try to connect he'll have that crapy message in server list 18:14 VanessaE um, the list is there prior to connecting to a server, derp 18:14 sapier just because you don't want to make your server tell him on connect 18:14 sfan5 1) why is that message crappy? 2) Not if they use an up-to-date verion 18:15 sfan5 version* 18:15 VanessaE update client --> no more message 18:15 sapier because it's a server list and not a "some message for all users" box 18:15 sfan5 how do you plan to inform users _in the main menu_ in 0.41? 18:15 sfan5 0.4.1* 18:16 sfan5 (w/o using the server list) 18:16 sapier did anyone of you even test how clients would react to that message? 18:16 sfan5 react? 18:16 sfan5 what message? 18:16 sapier if you abuse server list? 18:16 VanessaE wat 18:16 VanessaE it's just servers in the list 18:16 sfan5 the client reacts by executing some code and displaying it 18:17 sapier you add invalid data in there ... if unexpectedly someone added a check there or did some evaluation in worst case you cause a crash 18:17 sfan5 uh 18:17 sapier did you check it or not? 18:17 VanessaE three or four server entries at the top of the list sent to the client that when displayed spell out an "upgrade recommended" notice in the list 18:17 VanessaE invlaid data wat? 18:17 VanessaE invalid* 18:17 sfan5 invalid data != fake data 18:18 sfan5 you mixed something 18:18 VanessaE using the servers' descriptions is not invalid 18:18 sfan5 and someone adding a check.. uh 18:18 sfan5 maybe the dev team knows about this 18:18 sapier did anyone of you even test it? 18:18 VanessaE as long as the data exactly matches the specification the clients expect, then it's not invalid 18:18 sfan5 why would we need to? 18:18 VanessaE how can we test what you and PilzAdam vetoed? 18:18 sfan5 why do we even need to test this? 18:19 sapier serverlist is configurable you can make it download anything 18:19 VanessaE sapier: so you're saying that every server that ever joins the list requires every client to be tested? 18:19 sapier sfan5: because you suggest to abuse it and claim old clients wont be harmed 18:19 VanessaE sapier: to clients, this would look like four new servers have connected. THAT IS ALL. 18:19 sfan5 sapier: adding fake data != abuse 18:19 sfan5 and if you say it doesn't work, how about proving that? 18:20 sapier sfan5: in worst case that change might cause old clients to crash immediatly (i don't expect it but it is possible) 18:20 sfan5 yes 18:20 sfan5 if they crash on well-formed data 18:20 sfan5 thats not our fault 18:20 sapier YOU want this abuse so it's up to you to proove it's not wrong 18:20 sfan5 no 18:20 VanessaE prove it's not wrong? 18:20 VanessaE er... 18:20 sapier what's ip address of "Update to" ? 18:20 sfan5 sapier: 0.0.0.0 / 127.0.0.1 18:21 luizrpgluiz hi 18:21 sfan5 if the client crashes on seeing 127.0.0.1 in the serverlist 18:21 sfan5 that would be just plain stupid 18:21 sapier proove it 18:21 sfan5 no, you prove it 18:21 sfan5 seriously 18:21 sfan5 how the fuck do you thing the client is magically going to crash if it sees 4 new servers on the list? 18:21 sfan5 tellme 18:22 sapier I will not continue this discussion unless someone provides new information and better reason to do it till that I'll not agree as PilzAdam didn't 18:22 sfan5 oh 18:22 sfan5 you refuse to provide arguments 18:22 sfan5 can I see this as deafeat? 18:22 sfan5 defeat? 18:23 VanessaE sapier: you are the one asserting it will break clients. 18:23 VanessaE sfan5: implement it. sapier: prove it doesn't work. 18:23 sapier I provided all my arguments and did honor yours, result beeing your's not changing my mind 18:23 VanessaE basic science. 18:23 sapier ok noone did proove my recent patches don't work so I'm gonna merge them 18:23 VanessaE "Extraordinary claims require extraordinary evidence" 18:23 sfan5 sapier: please prove that the client will crash from seeing 4 new ordinary servers 18:23 sfan5 sapier: thats not how it owrks 18:23 sfan5 works* 18:24 sapier I don't have to provide anything to not agree to abusing server list as messagebox 18:24 sfan5 you do 18:24 VanessaE sfan5: the REAL trick would be inserting objects into the users' "favorites" list ';) 18:25 * VanessaE hides 18:25 sfan5 you have to prove that the client will crash on sight of ordinary servers 18:25 sfan5 VanessaE: lol 18:25 VanessaE sapier: you made the assertion. science requires YOU to prove your assertion. 18:25 sapier what does client crash have to do with you abusing server list? it's only one possible no go 18:25 sfan5 uh 18:25 sfan5 you brough the argument up 18:25 VanessaE wait, now the clients won't crash? 18:26 sfan5 why are you asking us how that has something to do with each other? 18:26 sfan5 you add invalid data in there ... if unexpectedly someone added a check there or did some evaluation in worst case you cause a crash 18:26 sapier I claim this might cause a crash as of science this theory is correct until you prove it to be false 18:26 sfan5 lolno 18:26 VanessaE kahrl: smartass ;) 18:26 sfan5 it is incorrect until you prove it 18:26 VanessaE sapier: ^^^^ did that quit message kill your IRC client? :P 18:26 sfan5 and a theory is not correct 18:27 sfan5 a theory is a theory 18:27 VanessaE sapier: that's not how science works 18:27 VanessaE you form a hypothesis and YOU test it 18:27 sfan5 a theory becomes a when it's true 18:27 sfan5 oh 18:27 sfan5 "fact" 18:27 sfan5 maybe 18:27 VanessaE perhaps :) 18:28 luizrpgluiz excuse me, sorry I dropped in a parachute, but this discussion leads to the minetest or to the IRC server? 18:28 sapier I still won't agree no matter how often you repeat your wrong arguments 18:28 VanessaE hypothesis: clients will crash. theory: bad data causes function X to return Y which causes function Z to segfault. Unable to confirm hypothesis. theory discarded. 18:28 VanessaE THAT is basic science. 18:28 sfan5 wrong arguments? 18:28 sfan5 which arguments are wrong and why? 18:28 sfan5 can you indicate that? 18:29 sapier "All clients have to update to use servers" <<< that's wrong 18:29 sfan5 who said they have to? 18:29 sfan5 We only say that they get a notice 18:29 VanessaE no one said they had to update to use the server list. 18:29 sapier if this isn't true then why damn abuse the serverlist? 18:29 VanessaE SCROLL DOWN A BIT 18:29 sfan5 nobody prevents them from using the server list 18:31 sapier You don't even provide a solution for future versions but only a hacky workaround for old ones 18:31 VanessaE [08-19 13:50] server list reads: Please update, your / client is outdated / 0.4.10 is available from / www.minetest.net / ----- SCROLL DOWN TO HERE -------- / / / etc etc etc 18:31 sapier VanessaE: and this is gonna be all you see on some android clients ... even with latest client ... great picture 18:31 VanessaE fAIL 18:31 VanessaE with latest client: no messages 18:32 VanessaE only real servers. 18:32 sapier if you add it to serverlist everyone will see it 18:32 VanessaE where "latest client" = Buildcraft 2 or your Android client (both are 0.4.10) 18:32 VanessaE NOP! 18:32 sfan5 heard of parsing user agents? 18:32 VanessaE NO! 18:32 VanessaE only OLD CLIENTS WILL SEE THE MESSAGES! 18:32 VanessaE jesus fuck 18:33 sapier Still why abuse serverlist 18:33 sfan5 because you can't get it into the mainmenu of the older clients any other way? 18:33 sapier provide a real solution first and then we can talk about hacky workarounds 18:33 sfan5 that is a solution 18:33 sfan5 but only for old clients 18:33 sfan5 we can worry about a solution for new clients later 18:34 sfan5 it's still time until 0.4.12 18:34 sapier If you believe this to be a solution why don't you get a majority of core devs? 18:34 sfan5 not everyone has commented 18:34 sfan5 and maybe because you do not understand it all 18:34 sapier then make them comment and get a majority 18:34 sfan5 make them comment, what? 18:35 sfan5 please tell me how to make people comment? 18:35 sfan5 s/?/ 18:35 sfan5 / 18:35 sapier You'll not get my agreement get your ok from others 18:35 sfan5 soo.. is "you didn't get agreement from others" an ok argument? 18:36 sfan5 if so, please remove all of your android stuff from the repo 18:36 sfan5 you didn't get agreement from everyone 18:36 sfan5 see the problem with your "argument"? 18:36 sapier right now you've got 2:2 18:37 sfan5 VanessaE is not a core dev 18:37 sapier 2 explicit disagreements against 2 agreements 18:37 sfan5 1:2 18:37 sapier even wors 18:37 sfan5 lolwat 18:37 sapier still I'd consider vanessaE close to core dev so I'd accept her opinion 18:38 sfan5 anyway 18:38 sfan5 what (real) arguments are left against it? 18:38 sapier Imho abusing the serverlist as message box is wrong ... especially if you don't provide a solution for future versios 18:38 sapier versions 18:39 sfan5 <sfan5> we can worry about a solution for new clients later 18:39 sfan5 <sfan5> it's still time until 0.4.12 18:39 sfan5 the biggest problem for servers owners now is that people are using very outdated versions 18:39 sapier sfan5: I don't believe you to implement that solution ... quite simple ... not for 0.4.11 not for 10.0 18:39 sfan5 s/for server owners/ 18:39 sfan5 / 18:39 sapier I believe once you've got that hack you'll never look at a real solution again 18:39 sfan5 sapier: thats not an argument 18:40 sapier you proved this with color table in past 18:40 sfan5 color table 18:40 sfan5 lol 18:40 sapier I don't forget things like that 18:40 sfan5 why do we need a color table? 18:40 sfan5 (need, not want) 18:40 sapier who needs that message? .. (20:40:29) sfan5: (need, not want) 18:41 sfan5 thats not an argument 18:41 sfan5 so you're out of arguments? 18:41 sfan5 guess we don't need a color table 18:41 sapier oh once I tell it it's not an argument but if you use it it is ... sorry with this rules I end this discussion now 18:42 sfan5 what is not an argument? 18:42 sfan5 mhm 18:42 sfan5 anyway 18:42 sfan5 this 18:42 sfan5 this 18:43 sfan5 is exactly why we never get anything done 18:43 sapier provide a solution then we can talk about a hack 18:43 sfan5 thats a solution 18:43 VanessaE any solution would require a protocol change...... 18:43 sfan5 a limited solution 18:43 sfan5 not a hack 18:43 VanessaE a proper solution that is. 18:44 VanessaE or not. 18:44 sapier VanessaE: Simplest way for a real solution would just download a "latest news" file and show it 18:44 VanessaE it would at least require clients to check the minetest.net website for updates anyway 18:45 Calinou we could have in-game news display too 18:45 Calinou if we have version check, that's possible 18:45 Calinou this is important for keeping community around 18:45 sfan5 PilzAdam: have you been watching the discussion? 18:45 sapier I don't think anyone sane would've followed that discussion ;-) 18:46 PilzAdam only the "but you have to prove!" part 18:46 PilzAdam quite amusing 18:46 VanessaE sapier: then 90% of this channel would follow :) 18:46 sapier PilzAdam: only because of not beeing part of it ;-P 18:46 sfan5 PilzAdam: not changed your mind? 18:47 PilzAdam I still don't like making the serverlist less usable by "blocking" it 18:47 VanessaE PilzAdam: not blocking. 18:47 VanessaE PilzAdam: amending. 18:47 VanessaE the rest of the list is still there. just scroll down. 18:47 sapier nice euphemism 18:48 VanessaE hell the top couple of real entries will still be visible on most devices won't they? 18:48 sfan5 PilzAdam: unless you use outdated version it will not affect you 18:49 VanessaE buildcrap looks to have about 5 entries' worth of space in its list, from checking screenshots on the google 18:49 VanessaE in the window I mean 18:50 PilzAdam sfan5, I am aware of that 18:52 sapier https://github.com/minetest/minetest/pull/1557 suggestions? 18:53 sapier sfan5: all of your issues fixed in there? 18:54 PilzAdam core.is_yes(core.setting_get("mod_profiling")) <- isnt there setting_getbool()? 18:54 sfan5 looking at it right now 18:54 sfan5 hm 18:54 sfan5 yea 18:54 sfan5 there is a setting_getbool 18:54 sfan5 and that should be used 18:55 sapier nice when was that one added? 18:55 sfan5 hm 18:55 PilzAdam its there since... ever? 18:55 sfan5 yeah 18:55 sapier hmm but no set_bool 18:56 sapier guess whoever did add this never heared about api symmetry 18:56 PilzAdam minetest.get_time_us() ^ get time in us <- at least spell it "microseconds" in the description 18:56 PilzAdam some people might think you can get the time in the united states 18:57 sapier ok 18:57 PilzAdam also can you filter the output by mod? 18:57 PilzAdam e.g. /save_mod_profile mobf 18:58 sapier not yet 18:59 sapier but should be a minor change 18:59 sapier I'll add that feature 18:59 PilzAdam yea, can be done later 18:59 sapier should be only a couple of lines 18:59 PilzAdam what does "min %" mean? percentage of what? 18:59 sapier of a server step 18:59 sapier I'll improve doc 19:00 sapier -doc + output 19:00 PilzAdam also it's "type" in english, not "typ" 19:01 PilzAdam everything else seems good 19:12 sapier you guys know that get_bool is only available for other settings file don't you? 19:13 sapier not for our main settings file ;-) 19:16 PilzAdam hm? 19:17 sapier core.setting_get() works, but there ain't a core.setting_get_bool() ... or is there a way to get the object of our internal settings? 19:17 PilzAdam its setting_getbool() 19:17 PilzAdam without the second _ 19:18 PilzAdam and there is also a minetest.setting_setbool() 19:18 sapier crap who did design lua api? 19:18 sapier I don't use the outdated "minetest" prefix in new code 19:18 PilzAdam yea, I was copying from lua_api.txt 19:18 sapier we've got get_pos getpos getbool get_bool ... 19:19 sapier all with "depends where you are right now" 19:21 sapier PilzAdam we need to fix lua_api.txt .. at least I don't find things in there any more ... even if I know it's in there 19:22 sapier you can't expect some newbie to use it if not even ourselfs can really use it any longer ;-) 19:23 PilzAdam you can split it into several files, but please keep the current version too 19:23 VanessaE don't split it, that'll just make searching it even harder. 19:23 sapier I don't think splitting alone would help 19:27 VanessaE if you can't find something in the API, either your search terms are bad (not very likely), or the API is worded badly. 19:27 VanessaE I would guess quite strongly the latter. 19:27 VanessaE splitting the API will solve neither of those problems. 19:27 VanessaE and only make the result of both problems worse. 19:28 VanessaE because then you have to try to remember "wait, was mintest.foo() in lua_api_visuals.txt, or was it in lua_api_graphics.txt, or...?" 19:28 sapier yet the doc is a open issue to me the current state is ... well ... not good 19:28 VanessaE I agree 19:28 VanessaE and you know I've had my opinions of it. 19:28 VanessaE the grammar sucks, it's badly organized, it lacks good examples, ... 19:28 VanessaE but when I've complained, people point at that Wiki site instead 19:28 VanessaE as if that's particularly good substitute for a searchable txt file 19:29 sapier I hate that damn wiki ... maybe because of that silly kitten thing ... but only because of it 19:29 sapier +not 19:29 VanessaE the wiki has too many circular references. 19:30 VanessaE I think I'm clicking on a link for more info, all I get is the same thing I'd find in lua_api.txt, and links from the page I ended up on take me back to where I was already, usually. 19:31 PilzAdam VanessaE, examples shouldn't be in lua_api.txt 19:32 PilzAdam it only bloats the file 19:32 VanessaE PilzAdam: there are many places in the lua_api.txt that don't even explain what the *output* of a given function is 19:32 VanessaE and no, I can't list them now 19:32 VanessaE I just run into them every so often 19:33 sapier can we add a rule "you may not mention coding style issues in lines not changed in a follow up review"? 19:33 VanessaE and yes, there should be examples, even if an example is one line. 19:33 PilzAdam what has this to do with examples? 19:33 VanessaE PilzAdam: because an example is directly related to the function's output? table in -> table out, two sets of coords in --> single item out? 19:34 Zefram_Fysh examples can be a useful way of explaining an API, and can to some extent make up for poor expository text. but I'd rather have good exposition to start with 19:49 sapier sfan5: suggestions how to do a better initialization? 20:00 sapier https://github.com/minetest/minetest/pull/1557 pushing in a few minutes 20:04 Zefram_Fysh cool! 20:16 sapier last chance to vetoß 20:16 sapier -ß+? 20:16 casimir Tested is just now. No problems. 20:17 casimir And I realized how consuming the leavedecay abm is. 20:17 sfan5 sapier: read my comments 20:18 sapier sfan5: I already fixed them if you didn't add some the last 10s 20:18 sapier btw I just merged it 20:18 sfan5 ik 20:18 sfan5 and that wasn't in the last seconds 20:18 sfan5 and you can rewrite history for 10 minutes 20:19 sapier which one of your comments was not honored? 20:19 sapier ah those you wrote while I was preparing for merge 20:20 sapier you do keep som issues back till I fixed the other ones on purpose don't you? 20:21 sapier ok rewrote history done now? 20:24 sapier https://github.com/minetest/minetest/pull/640 any chance to get this in if I rebase it? 20:27 PilzAdam sapier, shouldn't it run from the top to find the actual surface instead of caves? 20:27 sapier well if there's a tree you'll find a tree 20:27 sapier so either way it's wrong some time 20:28 PilzAdam checking for a tree is easier than checking for a cave (in a mod) 20:28 PilzAdam and moving additional 5 - 6 nodes should be fast in mods 20:28 sapier well I can change it ... as I don't really care 20:29 sapier but trees are way more common then caves 20:29 PilzAdam or make it more generic and allow arbitrary direction (i.e. a step vector) and arbitrary attributes to be checked 20:30 PilzAdam (so one could search for dirt_with_grass) 20:30 sapier if you insist in making this a generic evaluation function I'll just gonna drop it 20:31 sapier I wont spend a day in making this a perfect solution for everything, there are lot of ppl out there needing this way more then me 20:31 PilzAdam it was just an idea 20:31 sapier I know ... it's always an idea (generally speaking) ... and in the end a simple feature becomes a quite complex task and ppl comlain about complexity ;-) 20:32 sapier I'm closing it it's not worth the effort required to get it in shape 20:32 sapier ... if other ppls features just would be evaluated same way .. 20:35 casimir Found that I have an quite old pull request open https://github.com/minetest/minetest/pull/875 20:37 sapier casimir: is it still valid? 20:38 casimir Yes 20:39 sapier how to test? 20:40 casimir Place sand one node above grass and let it fall. 20:43 casimir It's not necessarily a bug, but it seems illogical when things get dropped that are supposed to be buildable to. 20:53 sapier even after reading this 10 times I don't understand what this is gonna be to change guess only way to find out is trying 20:58 sapier VanessaE: you're running https://github.com/minetest/minetest/pull/1475 for about a week now true? 21:03 casimir Currently when some node, like sand, falls and would land inside a node that is non walkable e.g. papyrus, the papyrus is dropped as item. 21:03 sapier I don't consider this to be that wrong 21:04 casimir When it lands inside a node that is buildable to it also dropps that as item. 21:04 sapier ah 21:04 sapier ok so first is correct but last isn't? 21:06 casimir Yes, it is more logical when the node is just removed. As this is the reason why there is buidable_to. 21:07 sapier sounds correct to me ... any other coredev same opinion? 22:21 VanessaE sapier: yes, I'm still running that patch 22:21 VanessaE not sure how long now, but at least since you had me initially try it, but I have not updated the server in a while. 22:21 sapier so your opinion good to merge? 22:21 VanessaE the commit in my local repo is dated Aug 8 22:22 VanessaE have there been any changes to it since then? 22:22 VanessaE (it looks the same to me) 22:23 sapier no I didn't touch that one in a while ... to busy doing crazy discussions ... again ;-) 22:24 VanessaE then I'd say it's good, but until those huge lag spikes in minetest_game are solved, it's hard to be 100% sure 22:24 VanessaE (assuming it is in fact minetest_game and not something lower-level) 22:25 sapier havent those spikes been before too? 22:25 VanessaE what? 22:26 VanessaE OH 22:26 VanessaE yes, they happened before this patch 22:26 VanessaE but until they're solved, it's hard to see the improvement 22:27 sapier actually you should see immediate improvement if you dig a lot of nodes for example 22:27 VanessaE yes, that much I've noticed 22:27 VanessaE though sometimes lag still gets the better of me 22:31 VanessaE regarding the surfaces discussion I missed, why not just take the function from plants_lib and rewrite it to C++ ? 22:31 VanessaE it's generic and elegant 22:31 sapier do it 22:31 VanessaE ha! 22:31 sapier I did this once did rebase it 3 times by now I wont do it again, quite simple 22:31 VanessaE (that was directed more at others than you really) 22:32 sapier I tried this version as I alread did have it available, it's not been accepted so I'm done with it now, as noone else usually cares chances that feature will be done are quite low now 22:33 VanessaE besides, yours is more of a "find the surface in this 1x1 column" function isn't it? mine finds whole, flat 2d(ish) surfaces. 22:35 sapier it's find surface at exactly this 2d position within a certain height range 22:35 sapier quite a simple task but doing it in lua spends a lot of time 22:35 VanessaE yeah, my code solves a different problem then :) 22:36 VanessaE yours is probably suitable for the job it was aimed at 22:36 VanessaE spawning players, animals, etc 22:37 VanessaE kinda surprised that didn't make it in yet, that IS a needed function. 22:38 sapier I hate that damn text overlaying formspecs 22:38 VanessaE yes 22:38 VanessaE that definitely needs fixed 22:38 sapier Zefram_Fysh: I've got a solution for our tab problem 22:38 VanessaE I have to field complaints of "pipeworks mese sorting tube buttons don't work!" or parts of Unified Inventory similarly 22:40 sapier it's quite ugly if your tab header is right wihin the "unclickable" area 22:41 VanessaE yeah 22:41 sapier some workarounds are quite ironic 22:41 sapier to make tabheaders work with variable fonts you have to disable them 22:41 VanessaE tabs at the bottom? /hides 22:44 sapier because if you disable the tabheader the event is passed to formspec menu and there you can detect it's actually been meant for tabheader, switch to correct font, enable tabheader pass event, disable tabheader revert font 22:45 VanessaE talk about hacky 22:47 sapier actually there's only two changes to forms left im my patch, button alignment and image size ... if I did change those back forms would look almost same as before 22:48 sapier same crapy positioning, same messed up positions on scaling ... but at least at default screen size they'd be same again 22:50 sapier well as there's noone having nuts enough to agree to fixing the formspec mess I'm gonne revert those positioning fixes 22:54 sapier hmm maybe I should've read ShadowNinja's comment first ... 22:57 VanessaE well you can always revert the revert :) 22:57 sapier usually not because I don't keep old experimental versions 22:58 VanessaE github does 22:58 VanessaE the code is still there. 22:58 VanessaE oh, you merged the profiler? 22:58 VanessaE \o/ 22:59 Zefram_Fysh re formspec, my patch is complete, I'm working on screenshots to demonstrate it. will be done soon 22:59 sapier I want the code not the screenshot ;-P 23:03 sapier VanessaE: where to find circular saw mod? 23:04 VanessaE it's in moreblocks 23:04 sapier ok how to show it's formspecs? 23:04 VanessaE in dreambuilder, you can just search "saw", grab one, place it, right click on it 23:05 VanessaE you can place wood or stone or a whole shitload of other stuff into the "Input" slot. 23:06 sapier how to place? 23:06 VanessaE just right click to place it like any other block 23:06 sapier ahh got chain saw not circular saw 23:06 VanessaE haha 23:06 VanessaE yeah, Sokomine should have called it a "table saw" rather than a circular saw 23:06 VanessaE (to me, the latter is hand-held) 23:07 sapier any other formspecs in dreambuilder? 23:07 VanessaE Color machine 23:07 VanessaE that one's pretty complex 23:07 VanessaE have some wood AND some wooden fences in your inventory to use for testing 23:07 VanessaE (it operates differently depending on whether you're handing it a craft item or a node) 23:07 sapier how to get it? 23:08 VanessaE search "color" I think 23:08 VanessaE :) 23:08 sapier spray booth? 23:08 VanessaE that's the one 23:08 sapier I hate enter leaving inventory 23:08 VanessaE "search words" -> tab -> space 23:09 VanessaE keyboard trick I figured out to avoid that. 23:09 VanessaE to see the more complex part of the formspec, "show supported blocks" 23:09 sapier ok color machine is completely messed up if you have scaling fonts 23:10 VanessaE eep 23:11 VanessaE also, technic CNC machine 23:11 VanessaE (search "cnc", looks like a grey spray booth) 23:11 VanessaE needs I think LV power to operate but you don't need to run it to see the formspec 23:11 sapier everythin relying on button to have fixed size ... wonder how messed up they already are on android 23:12 VanessaE *scratches head* what else have we got in DB that's reasonably "busy"... 23:22 VanessaE that's all I can think of at the moment :) 23:23 VanessaE unless you wanna test larger text entry forms. signs and LuaControllers that use those. 23:24 sapier spraybooth passes button height which was ignored by now ... that's why buttons get bigger 23:24 sapier at least that one could be fixed 23:24 VanessaE right. since their relative dimensions are fixed (or supposed to be), if you pass the height, the width should follow 23:24 sapier hmm ... no not completely 23:25 sapier button height was completely useless by no 23:25 sapier now 23:25 sapier guess making it useless again will cause less harm then fixing that bug 23:25 VanessaE heh 23:25 VanessaE well that's something that can be addressed later 23:26 sapier no it ain't as that is one of the major glitches 23:26 VanessaE incremental changes might be easier to adapt to than "holy fuck, everything broke all at once" :) 23:26 VanessaE oh. damn. 23:26 sapier I'll not touch it for sure later 23:26 sapier and as modder I'd prefere to fix my forms once instead of once per month 23:26 VanessaE true 23:27 VanessaE unless it's so utterly broken that you end up having to do a complete rewrite 23:28 * VanessaE hides behind the rubble known as "old version of plants_lib" ;-) 23:29 sapier colormachine even uses different button heights :-) funny 23:29 sapier noone ever complained about their buttons not resizing as requested 23:29 VanessaE the result of some hack to get around a bug in the formspec code probablyt 23:29 * VanessaE pokes Sokomine 23:29 VanessaE just about time for her to be getting out of bed now 23:30 sapier 1:30 am? 23:31 VanessaE oh that's right 23:31 VanessaE I keep forgetting she's over on your side of the pond, not down Under 23:32 VanessaE still, she's usually up around this time when it's hot outside. says it's usually cooler in the house at night 23:33 sapier hmm sorry vanessae but it's quite cold in germany atm 23:33 * VanessaE shrugs 23:34 VanessaE it's what she said the other day 23:35 sapier it's cold for about 2 weeks now 23:35 VanessaE hm, maybe not just "the other day" then :) 23:35 VanessaE *checks logs* 23:35 VanessaE shit that was almost a month ago now 23:35 VanessaE where'd the time go? 23:36 exio4 who knows! 23:36 VanessaE Aug 15 2014 23:27:54 yes. was very hot the past weeks. at night it was far better 23:36 VanessaE ok not a month ago, a few days. 23:36 VanessaE (looked at the wrong log) 23:36 VanessaE maybe it's just cold where you are? ;) 23:37 sapier it's cold everywhere in germany right now, 23:38 * VanessaE shrugs :) 23:38 VanessaE no matter 23:39 sapier color machine would be quite some work 23:40 sapier didn't Zefram_Fysh just want to do a few screenshots prior showing us his code? 23:40 VanessaE that's what he said but he's very meticulous about making things be *just* right 23:40 VanessaE so I give him plenty of space. he's *that* good. 23:54 Zefram_Fysh ready now. git://git.fysh.org/zefram/minetest/minetest_engine.git branch zefram/form_scale -- http://81.2.79.47:8989/gitweb/?p=zefram/minetest/minetest_engine.git;h=refs/heads/zefram/form_scale -- screenshots at http://www.fysh.org/~zefram/tmp/form_scale.html 23:56 Zefram_Fysh took me a while to sort out the screenshot toolchain. sorry to keep you all waiting 23:57 sapier I'm ignoring that IGUIEvent thingy as it's not required 23:57 Zefram_Fysh I fixed the tab-control font thing by creating a child of the tab control that intercepts its events and wraps OnEvent with setting the skin font. I'd like to compare that against the solution that you've mentioned 23:58 VanessaE Zefram_Fysh: wow, you've got your screen brightness seriously tweaked. anyway everything looks right to me. 23:59 Zefram_Fysh I've got inconsistent results about the gamma of those files. different tools displaying them at different brightness, and I'm not sure which is correct 23:59 VanessaE Zefram_Fysh: this is correct brightness, for reference: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2008172014%20-%2006%3a05%3a11%20PM.png