Time Nick Message 02:25 Guest54 celeron55 sfan i urge you to reconsider changing the default rendering for optionally framed glass. i found that a large amount glass mods uses that type, but it is quite often used together with glass planes or angled elements that are not framed. minetest game obsidian glass in particular is often used for the grid effect – but becomes 02:25 Guest54 invisible with connected glass. here are four visual comparisons that show how badly it affects existing builds near spawn on different servers: https://mister-muffin.de/p/C0Pf.png 02:27 Guest54 besides, it does not really work for texture packs with texture widths that are not multiples of 16 – even if the texture pack author does everything right (i.e. provides the additional texture so there is no fallback to mtg texture), the frame of the connected glass does not match, with does look jarring. 02:35 Guest54 btw, the connected glass option also changes the generated inventory icon, since the backfaces of it are also rendered while I think they are not with normal glass 02:44 MTDiscord Seems like the whole _optional thing is more of a landmine than anything. Mods that want a particular look should just use one or the other drawtypes. Anyone building with a material with optional looks must account for the possibility of either setting on viewers. 02:47 Guest54 btw, note also that it was rejected to make obsidian glass in mtg glasslike_framed for exactly the reason i am stating: existing builds use it for the frames. 02:48 Guest54 Warr1024 of course the option is a landmine. should probably never have been added. here someone brought up a particularly funny case of how it affects obsidian glass over a lava pit: https://github.com/minetest/minetest_game/issues/1547#issuecomment-335011698 02:55 Guest54 Warr1024 i suspect most builders don't really think about the optionally framed look (i certainly never did when i built stuff entirely out of glass) – but even if they were, glass mods just used the drawtype to provide the option to some. in mineclone2 for example, glass was changed to the optionally framed glass a year ago to cater to some 02:55 Guest54 users who like the look of the minecraft mod optifine (which does connected glass). most builds I know that use a lot of glass (e.g. stained glass church windows out of full glass nodes) are from before then. a big problem is that there is no way for modders or server owners to change it back to the old default. 02:56 Guest54 like, if you make it into glasslike, you can't provide the option anymore. 02:56 Guest54 anyways, it doesn't really work on glass panes or angled elements either, as you can see. 02:57 Guest54 oh, also a bunch of nodes “support” framed glass by … defining a transparent texture. ;) 03:04 MTDiscord framed panes when? 03:12 Guest54 Kimapr i believe there was an issue open for it, but i can't find it at the moment. 03:25 Guest54 so there is one thing i do not understand: given that several times before changing the default drawtype of optionally framed glass was discussed and always was not done, what exactly was the argument that suddenly made it acceptable to change the default rendering of optionally framed glass in very mod when it was previously agreed to be so 03:25 Guest54 disruptive to change a drawtype that minetest_game could not even change the node type to glasslike_framed? if i understand it correctly, the PR was only submitted after rubenwardy removed the “supported by core dev” label on the relevant issue (and whatever that may mean I do not know, the change is trivial, yet was never done despite supposed 03:25 Guest54 approval which was then … rescinded?): https://github.com/minetest/minetest/issues/8290 03:26 Guest54 very → every 11:06 MTDiscord Someone decided to do it 15:10 pgimeno Microsoft is Not Evil™ 15:18 Guest54 additional information regarding glasslike_framed_optional rendering: texture packs *larger* than 16×16px also break if connected glass is active, since the glass border needs to be *exactly* 1/16 of the base glass texture, or else the thing is going to look at least slightly weird 15:21 Desour Guest54: minetest has many issues, if you find more, open issues iff they're not duplicates. irc is not the right place for issuing issues, nor for providing more and more details in a long monologue. but I think you were told this already 15:22 Guest54 you can see this for yourself if you try minetest game with these texture packs and connected glass: handpainted (border looks fuzzy), sharpnet photo realism 64px (glass becomes invisible, frame has weird double borders and edges) 15:22 Guest54 Desour well i can't open an issue right now, but i will shut up regarding this, i believe i have provided enough information 15:24 Guest54 also my previous experience is that if something has been discussed on github and disregarded, clearly explaining it on IRC may make people who dismissed criticism look into it again, but i get it, it's the wrong place. 15:25 Desour there's also the forum, in case you don't want to / can use github 15:25 Desour writing into irc is also ok, but looks like it's starting to get out of hand again 15:27 rubenwardy honestly, reading that issue it feels more like the solution is to deprecate glasslike_framed_optional and not change the default value of the setting 15:42 Guest54 rubenwardy reading https://github.com/minetest/minetest/issues/11546 maybe a new drawtype glasslike_connected to decouple the 1/16 frames from the connected look is an option long-term ig 15:59 rubenwardy sfan5: users reporting broken models with latest irrlicht 16:02 Desour_ tell them to try with revision 10 (AFAIK, that's the one we want to ship with the next release, right?). 16:02 Desour_ https://github.com/minetest/irrlicht/tree/1.9.0mt10 16:03 rubenwardy either the engine should require 1.9.0mt10 or you shouldn't be pushing features during a feature freeze 16:07 MTDiscord Guest54: I created the PR before I was aware that the core dev support was removed. 16:07 Krock added the changelog for 5.7.0 (based on master branch from today) 16:08 Krock looks like MTG also got some updates. adding them as well now. 16:09 MTDiscord Naturally, the first time I try to contribute to the engine, the change becomes very controversial -_- 16:09 MTDiscord why did I even try. 16:11 rubenwardy I removed the label because there wasn't a clear consensus in the issue 16:11 Guest54 well, it was always controversial, you just were not aware of the history of it (or the rendering issues that basically mean your texture pack has to support connected glass stylistically) 16:11 Krock well, such problems that can happen to everyone 16:11 Guest54 it happens 16:12 Guest54 yeah 16:12 Desour_ s/can happen/do happen/ 16:12 Krock although I wonder why texture packs have no control about that... ? 16:13 Guest54 because it does not make sense rendering-wise? 16:13 Krock why not? texture packs are literally what's used to render 16:13 Desour_ +1 for per-node _optional decisions using texture packs 16:14 Krock or as mentioned previously - to remove the optional part 16:14 Krock I see that it only overcomplicates the matter 16:30 Guest54 PrairieWind to be clear, the “problems happening to anyone” is not a personal failing, but a failing of a system. humans know at least since the 1960s or so how to develop bug-free software (AFAIK the apollo guidance computer program used for the moon landing had no known bugs, decades later), but it requires *a lot* of effort – think of the 16:30 Guest54 most effort you could put into making sure a sorting algorithm works and then think what if every line was written by someone different as a hobby project … and then add some. almost no one does it, since developers usually value new features more than not breaking stuff (reasons: it is more fun, it makes economic sense, humans have a limited 16:30 Guest54 life span etc. pp.). 16:34 Guest54 PrairieWind in this case, you could have noticed the rendering issues or thought about the consequences more or tried one of the numerous texture packs or read the at least 4 past issues about this, but in the end someone approved it and presumably also did not do all that. you can not blame them as individuals either, as long as minetest is a 16:34 Guest54 project where developers and users alike prefer getting new stuff instead of getting a bug-free experience (as evidenced by shadows, which were once too buggy to be part of a release). 16:35 Guest54 so i hope i did not demotivate you 16:36 MTDiscord well, I guess I should stick to the supported by core dev issues, and asking first. 16:36 MTDiscord especially if they are older issues 16:36 Guest54 PrairieWind besides that, “let's make a release candidate and then wait for complaints” works well enough in practice (as evidenced by me complaining). 16:42 Guest54 PrairieWind IMO the easiest way to figure out if something is a sensible change or addition to a program is to go outside of your comfort zone. look at the implementation and think of how it could break. e.g. if you want to work on keyboard-related issues, ask people to test it on different machines or with a layout different than QWERTY etc. 16:44 Guest54 also if someone claims to develop bug-free stuff, ask them how they did it. usually “oh i just don't make any mistakes” means they are charlatans while people who can explain stuff like unit testing or input validation or static analysis are able to help you to become just as good as them. 16:45 MTDiscord speaking of that, I have an add favorite server button that I might need user testing and input for because of how the highlight thing is working. 16:46 MTDiscord if you click the button, it adds the server to the favorites list, and highlights the next server down. But it shows the favorited server information in the side panel still. 16:47 Guest54 why does it do that instead of keeping the highlight on the server that you added to the fav list? 16:47 Guest54 (which would be what i expect) 16:47 rubenwardy The server list selection should probably be based on address/port rather than index, would be more reliable 16:48 Guest54 oh i see, the index thing makes it “highlight is i positions down in this list” 16:51 appguru sfan5: re https://github.com/minetest/irrlicht/commit/edb381bd5050712d1eb8875fe3a405000dd09a3d: it is not apparent to me why flipping the sign of the W component of a quaternion representing a rotation should change the handedness 16:52 appguru I believe that instead the X, Y and Z components should have their signs flipped, since that would be akin to flipping the direction of the rotation axis which turns the rotation converts the rotation to its inverse (as if you would rotate counterclockwise rather than clockwise) 16:55 Arc_ is there any documentation that explains the lighting algorithm? 16:57 Guest54 Desour_ the security vulnerability here is not exploitable client-server-wise, or is it? https://github.com/minetest/minetest/pull/13315 16:58 Guest54 i mean, the client can't arbitrarily trigger it 16:58 Guest54 in any way, i suggest to issue a CVE because that is what makes linux distributions take note and backport the fix to earlier versions 16:58 Desour_ a client can only trigger it with the help of a server-side mod 16:59 Desour_ (by abusing that mod) 16:59 Desour_ so no 16:59 MTDiscord yah, checking by address and port would definitely be easier than idx 16:59 MTDiscord *index 17:00 MTDiscord the index is whats giving me an issue. 17:03 MTDiscord Minetest has existing CVE's that require a faulty mod to be present, such as the minetest.deserialize & ItemStack injection ones 17:06 Guest54 that last one, in fact, was exploited so thoroughly that you may still be able to find remnants of it on old-enough servers 17:08 Guest54 funnily enough, everyone i know who used it claimed to just create OP items with it (since rooting a server is outside of the game or so?) 17:27 Desour_ I haven't used the github security advisories thing yet. (I'm also not sure if I have the rights to create one.) it's not the first and certainly not the last UaF we'll ever have fixed. it's also not as trivial to exploit as the other things that got CVEs. it would probably still make sense to open one though, and in the future open one for every other tiny security-relevant issue 17:54 sfan5 rubenwardy: if set the release mode cmake will require exactly 1.9.0mt10 no earlier no later 17:54 sfan5 so irrlicht does not need a freeze 17:54 sfan5 @appguru I will revert the commit and then we can maybe send a bug report to the irrlicht dev 18:52 GuestBanana #13349 opened 5hrs ago, coincidentally I went looking for a feature request of that just now. Seems to be a duplicate of the #11362 which was closed by paramat for inactivity. Possiblity of re-opening it? 18:52 ShadowBot https://github.com/minetest/minetest/issues/13349 -- Defining Crack Animation Per Node 18:52 ShadowBot https://github.com/minetest/minetest/issues/11362 -- leave cracks while breaking a block 18:53 GuestBanana Sorry, that second one should be #3540 not 11362 18:53 ShadowBot https://github.com/minetest/minetest/issues/3540 -- Provide the possibility to provide a crack texture override per node. 20:59 pgimeno @appguru @sfan5 what commit? note that if a quaternion Q represents a rotation, -Q represents the very same rotation, hence why changing Q.w to -Q.w works to invert the sense of rotation 21:00 Desour_ pgimeno: https://github.com/minetest/irrlicht/commit/edb381bd5050712d1eb8875fe3a405000dd09a3d 21:01 pgimeno thanks, looking 21:05 Desour_ cmake question: why are we using for some compile options set() with CACHE and for some option()? 21:11 pgimeno I have to give up, Irrlicht's axis convention and choice of Eulers for some rotations is just too horrible for my aging mind