>>1020613 # the post erasing is the worst part, on meguca is pure cancer. 3000 niggers all typing on the same thread and people are replying to posts not even there. Or replying to future posts, etc. unless the site has died down since the last time i checked anyways
>>1020648 # the thing their /pol/ did with having just one general they used like a chatroom made it too fast for me to keep up with but the liveposting format itself doesn't seem too bad once the novelty wears off and everyone is used to how it works.
>>1020885 # The problem with lain wasn't the anime but the low effort /tech/ niggers that it attracted. It's just another badge that people use to signal each other, just like an arch linux rice, etc. Lain's "fanbase" didn't deserve to see it. It just embodies low effort signaling like UNIX faggotry on /tech/.
>>1020906 # Tbh it makes a lot of sense why trannies love lain, literally the embodiment of "no one knows im a faggot on the internet". The themes running through SEL could easily be taken as a sort of gender neutral thing, because of the whole avatar thing.
>>1022030 # no one splits up on livechat chans, megathreads are the only things posted on. once a seperate thread gets too big everyone just joins in on it to shitpost. Usually livechat chans end up with a core group of only-shitposters who kill every discussion. usually, anyways
Liveposting sucks with more than 10-15 people.
This thread is so much calmer than the /pol/ one. Amazin'!!
>>1018974 # hey lads sorry if it looks like I'm mentally handicapped but: 1) why are all images broken unless I manually open them? 2) got any other bunker you're going to? 3) Anyone from /ita/ in here? It feels like that scene from Gone With The Wind where thousands of bodies are down and you don't even know who's going to respond or not, with all these alternative chans
>>1029477 # the media is keeping it simple by not mentioning also the Ohio shooter because then 8chan won't be the ebil dark web anymore
>>1029273 # Ottimo You could've still used a VPN or some proxy server or something, bro Also I don't think it was anyone from the government as much as Postal Police's work when doing massive sweeping actions on sites thought of hosting illegal pornography. One of our guys tried to convince them otherwise but sadly they didn't say what exactly was wrong with the board.
So with 8chan being ded for... longer than expected. I'm seriously starting to wonder what will happen to demo day. I mean, we could do it in a place like this. But how many /agdg/'ers will be here? And how many 8/v shitposters will be here?
If 8chan isn't running on 8/8, will we delay demo day?
On that note, are there any other places 8ch/v shitposters went to? I don't feel like bothering with tor/p2p/darknet shit, and there doesn't appear to be anyone here but agdg-fags.
>>1038237 # I'm working on sky rendering right now. Hopefully we will still be able to have a demo day. I'm not sure what the best plan of action is. If 8chan manages to come back up a few days later, it would probably be a good idea to hold demo day later there, or maybe just hold it once on the actual day and once 8chan comes back? I don't know how many people are even here though. There's also the endchan /agdg/ but nobody has even posted in it, as well as some chatrooms that a few people from /agdg/ hang out on. It wouldn't be enough to just post demos when 90% of the people on here are spread out in different bunkers or just waiting it out.
>>1038701 # >It wouldn't be enough to just post demos when 90% of the people on here are spread out in different bunkers or just waiting it out. That brings up a good point. I think in the future we need to be better about saying where to go and what bunker to use.
>>1039146 # >>1039201 # I plan on posting my demo on demo day and then continuing to work, so when 8chan does come back up online I'll be able to post a better demo. That seems like the best way to go forwards, since in the worst case scenario (8chan doesn't come back online, or takes a long time to come back) not trying to post demos would mean demo day wouldn't really have happened...
>>1039598 # I'm just thinking, I don't know what my posts are going to be like, until i'm writing them. I know how to type... that isn't what makes my typing speed so slow.
>>1037059 # Get in the bone remover, Shinji >>1028412 # Webm related. It reminds me of when moot decided to crash /r9k/ with no survivors. I was pretty attached to the board at the time and he just let /soc/ and the other normalfags run rampant.
>>1039273 # Nice, good lighting is important and easily overlooked.
>>1053769 # That's probably because these fags like final fantasy xiv, and then devolved it into a 4/vg/ (or 8ch-thing-specific board) thread where the content matter of the subject matters nothing and it's just one big chatroom.
This is why general threads are banned, by the way.
>>1053765 # On the note of demo day, what's gonna be the plan? To repeat my suggestion over at endchan, I think we should at least postpone the demo day until we decide where we do it. Instead of fracturing fucking everywhere.
>>1053757 # How alive is 08chan? Are there any /agdg/ stragglers? I can't be arsed to set that shit up, not to even mention the security/privacy concerns.
>>1053772 # >How alive is 08chan? Are there any /agdg/ stragglers? 08chan is probably the most active bunker at the moment, though the most popular boards at the moment are /pol/ and /qresearch/. >I can't be arsed to set that shit up, not to even mention the security/privacy concerns. It's pretty easy to set up, but you should definitely go through TOR or a VPN as everything is in public logs, including your IP apparently. There's at least one CP board, so that should be muted right away.
>>1053757 # I've tried to join 08chan but the content.json won't download for some reason. I'm tempted to uninstall and try again.
>>1054240 # Here's an example of a noneuclidean sky texture placement, I have it set up so that the sky will still render in that place, even if that doesn't conform to the euclidean rules of the world around it. So this pillar looks like a window into the cube map.
>>1055603 # Skyportal usually means that it just acts as a portal for rendering to another location. Some games/engines (e.g. source) use this and create a low-poly skyportal map next to the main map, where the skyportal directs to.
>>1055608 # I know what you're talking about now. The way my rendering works at this time I wouldn't be able to place models into the skybox, but theoretically I could change it to allow that. At this time it's just a really basic way of masking off where the cube map gets rendered by changing the depth buffer settings a few times while rendering invisible polygons. Because of how it uses the Z buffer it wouldn't work for placing models, but it's totally possible to change how it works to use that kind of feature.
Anyone ever do JS game dev with the Phaser engine? How does it compare to something like Cocos 2D? Also, any recommendations for sprite/map editors?
>>1053769 # Because this place was dead and when it was first advertised on 4ch, /xivg/ made a small community here. They were really the only actual community to make a place in /v/
Well here is my demo, it manages to show off a lot of significant improvements to the technology, with sloped surfaces and sky rendering, but it still is as buggy as ever, if you want to crash this demo, you will probably succeed, the r_restart command in the menu isn't perfect... and unfortunately I didn't even start porting back to linux, I was going to do that but all of the time just, slipped away, if you are a linux user just , reply to me and I will work on that faster, i'm probably just going to do it tomorrow. So anyway this demo comes with a menu, and I redid some stuff in the manual, like updating the guide to all of the cvars (all of the ones that I want to document at least, some of them are truly broken or useless right now). Anyway give it a try, the levels are kind of simple in this one because of serious bugs in the map editor that I really was not able to fix in time for the demo day, and so as a result the complexity of the maps suffered. I'm sure by the next demo day these problems will be fixed, and I can have new, exciting problems, maybe related to a lighting model or something which is the next logical step from where I am now.
>>1056248 # Well, the results are pretty much in. We're doing this shit anyway. Normally the demo day thread gets posted a few hours from this time, depending on when someone gets off their ass. Some people already dumped their demo's here. I still kinda want a separate thread somewhere. I'll just casually work on the last bits of my demo whilst I let someone else take care of that.
>What is demo day? A seasonal community event in which your fellow nodev anons show off their vidya, and for ''you'' anons to do what you do best. >That is? Play our vidya, and tell us how we can improve, show us the untapped potential you see, tell us how much our game sucks (or doesn't, after you've played it of course), and ''why'' it does or doesn't suck.
__'''Obligatory Notices'''__ ''Please post criticism/bugs/etc in this thread.'' ''After'' this thread is culled, please post criticism/bugs/etc... wherever /agdg/ will migrate to.
>>1060312 # Not sure how much of these graffics are intended but I'll just go ahead and point out that this looks out of place. Assuming this is the not-bionicle game you posted last time, the game still feels like complete ass. Controls need to feel good. I could go into detail, but literally fucking everything is jank. At least you made the dodge slightly-less-jank to the point that you can make it feel good with proper animations / movement, but I would really prefer something like "press space to immediately dodge". This is especially important if you want to apply iframes to dodges.
Also the camera inversion is the wrong way around. If it's inverted, it works normally. On top of that the vertical camera movement is incredibly sluggish. I can imagine it being good enough for controllers, but for mouselook it feels terrible.
>>1062287 # The terminal eats my inputs and does weird things with the input string if I type too quickly! The cursor moves around; pressing buttons too quickly makes it replace the last character in the string... The fuck is thing doing? With this kind of fuckery I expect it to crash.
Besides that the game runs & plays fine. You can bunnyhop though, which you probably don't want in a tactics game. I can press F to interact... but there doesn't appear to be anything to interact with? If you're actually going for the low-res thing, make sure that the text is properly readable. The characters are a bit too compact for that right now.
>>1062596 # >>1060312 # A question I do have to ask you is: What you're trying to achieve with your project? I'm not really into the more hardcore deep rpg stuff, and I've never played Gothic, so I'm not an expert on that subject. But if you allow users to directly control an entity, they will expect it to actually respond well to the inputs. If I could just rightclick an enemy and my character would run up and hit it, then all of this jank stops being jank and starts being gameplay mechanics. But the fact that you can actually dodge out of your own volition makes me believe you're trying to make this partially an action game. Which is why I will not stop calling your game janky garbage until it stops being jank as fuck.
I don't want to come across as an overly negative asshole multiple demo days in a row. But all I see is an action rpg that has the worst action I've ever seen.
>>1053925 # i couldn't get the link from the pasta to work either but there's one on the infinity board on miichan that worked . 08chan is steadily gaining more traffic but everyone's still kinda confused on how to use it and seeding is somewhat spotty. i think maybe that could be solved if i got a little home server or something just for /agdg/ and maybe an art board too. obviously the biggest hurdle is figuring out how to keep it moderated when (((they))) inevitably try and shut it down by spamming cp everywhere. >>1062715 # tired sry, what link do you mean?
I don't know what you plan to do with this game, whether it's going to be an arena fighting game, or a level/chapter game with drop-in/drop-out multiplayer, or if it's going to be more dungeon-oriented, but I'd certainly play more of it. I think you're doing good. I would have stayed around longer if I didn't have morning stuff to get done. If I have time later, I'll screw around with the level editor some. Keep up the good work, anon.
>>1062909 # The log file says the engine shut down correctly, so maybe you hit F10 or some other hotkey by accident that kills the application? I don't know the cause. If the engine does uninitialization, then that means it did go through a valid codepath to shut down. Which technically isn't a crash. There are some other cases in which the engine shuts itself down though, so I'll double-check those. >I don't know what you plan to do with this game The ideaguy in me wants to do fucking everything. But I decided to make the gameplay work first. Thank you for the compliments.
>>1062944 # That's true. All the RPG mechanics seem to work fine.
>>1061366 # The frame rate is oddly low for what this is. I had the program running at around 35 fps the whole time, using OpenGL 4.5.0. I also don't like that you can see through walls if you shove the camera into them, and that the alt key freezes the game until you press it again, but I'll assume those are bugs you're working on, so whatever. That aside, it worked fine for me.
>>1063693 # That's probably a bug in the framerate limiter. If you increase the value of r_fps to 120 you should get 60 FPS, I don't know why it happens exactly because it depends on the specific computer it's being run on and i'm not sure how to detect it yet.
I'm not sure what the problem with the ALT key is... but i'll try and figure it out. Being able to clip through the walls isn't really a bug, but rather a consequence of the hitbox being nothing more than a single point right now, which lets you get close enough to walls to intersect the near clipping plane with them. Eventually i'll program an actual collision detection system with real hitboxes.
>>1062632 # That's got to be a precision issue. Unfortunately arches3 still has some PVS bugs if you can find them.
>>1064130 # The engine is really fast, I can get 12000FPS at the most on my OpenGL 4.5 card, right now I am stuck on this intel integrated graphics which can only get 300 frames out of it, it's kind of frustrating because that specific bug has been in the engine in some form for years now, the problem is something with the QueryPerformanceFrequency() windows function that tells you how many ticks are in a second, it's like windows cuts this value in half or something but only on some computers and it's stupid. I bet that i'm doing something weird or something dumb is happening with it, maybe it's a precision issue and one of the bits is being eating, cutting the value in half..
>>1062705 # >>1062596 # here's the updated version https://drive.google.com/open?id=12C2kAYEUQLhx4J9yGmA-pjMm4grGk2z6 dodging starts almost immediately. if you still think it's too slow, i probably need to do something other than double tapping palm leaf material is fixed, as far as i can see the white outline is from the smoothness on the materials. i've lowered it but it is supposed to be there to some degree. if you're still getting it, you might have to turn anti aliasing on
>>1064304 # The dodge is definitely more responsive, probably because it just starts with faster movement. That said, the hotkeys to trigger dodge are not responsive. If I double-tap, it sometimes just doesn't do the dodge. The inputs regarding letting go of movement keys also aren't responsive. If I'm walking and let go, the character keeps walking for another split second. (Remember how last time I told you it felt like it was pathfinding based on inputs every x sec and told you not to do that? It still feels like you're doing that, and it feels like shit.)
I'm running it on a different machine now, but I'll reply again when the first machine I used nets different results later. The palm leaf material is fixed. The shimmering is greatly decreased; I assume you just toned down the white you blend to, and that is definitely the correct solution for this. Ideally you want to adjust the slope you blend on so that the sides of the models don't produce much brighter results than their adjacent pixels. I didn't need to turn on AA, but I decided to mess with it. Turning on the last form of AA, TAA, did cause visual issuse. The foilage starts to appear to "vibrate".
On the note of AA, your game does some fucky stuff with settings and it sometimes doesn't do AA either. >start exe >set resolution to 1080p, fullscreen, hit start >after a fullscreen loading screen, the game starts in a 320x240 windowed mode, have to mess with ingame menu to fix it >turning on AA doesn't actually produce an anti-aliasing effect (but TAA causes foilage vibration) and when I close the game and restart it, it does appear to save my settings, but it boots the game in windowed mode. In this case, AA DOES produce an anti-aliasing effect, and TAA produces a blurred foilage effect instead of vibration. But if I put the game back in fullscreen, the AA no longer functions, and putting it back to windowed doesn't fix it. Another restart does.
Some fucky shit is going on with your settings lad.
>>1064919 # TAA is kind of just there because it exists. It's meant to be used for moving objects, looks wonky on static stuff
>>1064919 # I should probably add here that a lot of modern games do actually have the character keep walking briefly after letting go of inputs, but play some sort of "brake" animation. That way it wouldn't feel as janky. (But still janky and you shouldn't do that.) (And blending the camera position instead of snapping it to the character also reduces jank here.)
>>1062287 # This looks like it could be really fun once there's some content. I have a few complaints about the terminal. The s key behaves weirdly and doesn't let you type fast. The first line of output is partially cut, while there's some extra space at the bottom, and that looks bad. And I wish the terminal would be larger than 8 lines, although that might be fine, depending on what you do with the terminal in the game.
>>1065112 # >>1064919 # huh, apparently changing the fullscreen/windowed mode fucks the postprocessing. i have to disable/enable it a frame later and it works again. go figure
>>1065112 # >>1064919 # reuploaded it, same link i think i fixed the issue with the settings. for the movement i lowered the time for it to set the speed in the animator, but it still can't be instant, otherwise you get situations where you switch directions and your dude does a 180 instantly
>>1062721 # The controls are fine. I found the impact of the hits a bit underwhelming, although I can't say it isn't there. There's no way the attacks are gonna feel right until you make the animation match the hit box though. I dislike how you take a step with every attack, especially with how short the reach is. The buffs need some visual and/or sound effects, at least on activation. The combat is too shallow right now, but I don't think you need us to elaborate on that.
I still have to fix text characters disappearing, it is an issue caused by holding/pressing any keys bound to Hotkey1/2 or Command1. What is supposed to happen is if you just press the input then text is inserted, holding should make the terminal's text scroll faster when holding Up or Down, the issue is that queued input often gets deleted due to trash scripting. Terminal is going to be kept at how many lines it has, though I agree that there is an issue where the bottom line is cut off and it shouldn't be doing that. You can scroll down to it using Down though. And the text is too small, I agree. This will be fixed, lower case "s" is especially hard to make out (font size is currently 8). I've been putting together other elements of the game so I haven't gotten around to fixing these issues yet but should be fixed in v0.0.2.
Bunny hopping stays in, I really enjoy it. It is movement tech I guess, it is caused by the player having a high enough air control variable that it causes them to move faster than their max run speed while in the air. This isn't a bug, it can be fixed by reducing the player's velocity limit. Also there is nothing to interact with in this level, that will be in the next demo.
Low resolution fixed, went from 320x240 to 640x480 because my sprites get awfully blurred the farther away my camera is.
That said, thanks for testing it out. There isn't much to it right now, I just wanted feedback on player movement and readability. Lots of the same feedback so I ikn
>>1072267 # Whoops, hit enter early. Lots of the same feedback so I know what needs to be updated. For everyone else, I'm going to play as many demos as I can get my computer to run tomorrow, slept most of today (had 3 hours of sleep yesterday).
>>1061366 # Tried your demo last night but I couldn't get the UI to work. Enter and Space does nothing to choose any of the elements on the main UI screen so I'm guessing I'm doing something wrong. It seems to run at a decent frame rate though. Terminal has issues, input is passed to it even when hidden or if I have a different window focused other than the game window.
>>1072524 # Do you have a non-US keyboard? It's supposed to respond to the spacebar. Did you manage to open any of the maps? You can use the launch argument -map or "map" in the console with the name of the map you want to open , like "sky4_dd3.sw4" or something
>>1074249 # The steam version is 100% different. This one does not have the fairy village thingy and the cooking tries to be more like atelier-like crafting rather than some gay simulation Unfortunately, the many numbers still have no meaning, but we are going to eventually implement some bonuses like running speed and energy conservation if you can keep a nutrient balance. The tastes thingy is an arbitrary grading to tell how good the food item you made is (The grading is still not implemented, but you can see what each recipe wants from your item). The yellow tag on the item is the aroma, when implemented, the girls will complain or like if the aroma is something they like/dislike, etc
>>1020574 # this. I want all the electionfags to fuck off so we can have a real /pol/ again
>>1020688 # then its not a board, it's a chatroom. Goes use discuck
>>1062721 # Gonna take down the server now. Many thank to all the people that dropped in.
>>1083589 # yeah, apparently you will not download any of that board content unless you actually click on it, just like your browser cache you only host sites that you have visited, and only as much as you have visited yourself, so you won't be hosting any of /pol/ content unless you actually go to /pol/ or /cp/ for that matter
>>1084305 # The bigger issue is some faggot actually going through every single board and thread and spamming CP, as it's very hard to actually ban someone, rather, it's impossible But then again, it's the same issue of your browser cache now having CP
I wish I could sage here when I forget to write something and end up double posting
Pulseaudio is such a piece of shit and has awful documentation. I'm working on the audio for my engine and have decided to just use ALSA and maybe OSS chosen at compile time instead of runtime. I see why Sigma engine uses OpenAL this is a headache.
>>1084945 # To be fair, I didn't even support Linux. But sound is a problem that I have to solve again, unfortunately... and I can't just take shortcuts (making people run oalinst.exe is unacceptable to me). I might support OpenAL just to future proof the engine. If you want it to work, I could force people to put the dynamic library somewhere and hook into it if it's there.. that could be interesting.
>>1084945 # The reason why OpenAL is so widely used is because it "just werks". Same thing with SDL. I'm sure it has plenty of shortcomings too, but they don't matter when you're getting started.
apparently this whole time my directional light was yellow instead of white. better or worse?
I didn't realize that Carmack didn't write the audio for doom. I was looking at Doom and Quake for inspiration, but both of them are before the modern (read as shit) linux sound ecosystem. I was surprised Doom had a separate process running sound on Linux. I would have thought that would be a lot of overhead at the time. ALSA wasn't chosen on technical merit, but because of OSS going proprietary. Every other free unix like just built off the old free version of OSS, but Linux fags had to have a special snowflake API.
>>1096746 # Is Mark still storing unhashed IP's like a retard?
at any higher than the bare minimum filtering settings for the lighting with progressive GPU, my whole computer freezes. with the bare minimum you can still see a fair bit of noise in the darker areas. however with CPU i can easily do the thing at any setting, but it takes like 4 hours wat do
>>1085450 # Really cool but makes me dizzy after a while
>>1120532 # Sigma 1 had either lightmaps or distance-based shading. The lightmaps were done in this hilariously inefficient way with NO multitexturing at all, and effectively doubled the polycount because it would redraw the scene twice with the Z buffer turned off (this was doable because it was a BSP tree. You don't need the Z buffer to draw a BSP tree because it's in order). Combined with the lack of any culling whatsoever it was just very slow.
Any way right now Sigma 2 has no shading options at all, this is because i've just been doing work on other parts of the rendering engine, and so I haven't gotten to lighting yet. However that's probably the next feature I will be working on.
Disgusting crosspost. I laughed my ass off while making this a few weeks back
Audio now works on linux with ALSA. I have no idea what I changed to make it start working though. Something I found was that the default buffer has about twenty seconds worth of audio so you have to play with the settings to find a small and supported buffer size. Now that I have sound, text, and graphics I can either work on the Windows port or make a game instead of a model viewer with a frame time counter.
>>1150115 # oh boy getting Chaos Theory vibes with this minimap
Put it on the watch, not sure if this or a floating holograph view like Metriod Prime/Dead Space would be better/cooler. The map is harder to read on the watch but I feel like maps shouldn't give everything away, or you lose an exploration aspect to the game.
just have it rotate independently from look rotation
I was searching for ideas and stumbled onto the scratch forum. They have dedicated section for ideas. Does it makes sense to filter them and add interesting ones to the wiki? https://scratch.mit.edu/discuss/9/
Assume that you have a regular grid for movement, but also have a fancy navmesh made from planes with 4 corners. How would you always know if and where you're crossing an edge on the navmesh, even when you're right on top of an edge or moving parallel to it? Because right now those sitiations are fucking me up due to floating point errors. In the picture all of the blue arrows are valid traces, but if they collide with a red edge they'd be invalid
>>1168527 # I might be wrong, but how is that grid for movement? As I understand, grid for movement means that you can only move between the nodes of the grid. Isn't your task just to find out if segments are intersecting?
>>1169963 # the grid has weights for stuff like flocking, ignore it, the passable areas are the issue. I've got one dude who's arleady created a path and everyone else is following him. they each trace towards their target position to know if they've entered a new area, or to know if they're about to cross an impassable (red) edge the problem is that if they're right on top of an edge, or parallel to it, they can't properly figure out where they intersect with it and as a result they end up missing the moment when they're supposed to cross over to a new area
>>1169967 # Oh, I finally understood the issue. You standing exactly on the wall an you are moving in parallel to the wall and walking through the wall due to the error. I'm not an expert, but I feel like you shouldn't let your actors get that close. Physical objects have value. So, it's reasonable for points to have margins.
>>1169976 # i've already got a number of fallbacks to try and prevent that, like avoiding the corners or if failing to enter an area check if you're currently in your designated areas. but even then, there's nothing stopping you from walking parallel on a passable edge. in most cases i can then check if the neighbour areas contain the position, but when the mesh becomes small and complex this still ends up missing some cases where there's a lot of small areas.
>>1169982 # I think if you used circles for actors instead of points it would be easy to detect if it's started entering, entered or outside. If actor attempts to enter via illegal edge you just throw him outside. Yeah, no physics involved, but principles are the same, aren't they?
>>1170004 # the dumb solution right now is to avoid complex/small meshes. The whole reason why i ended up asking was because i had to make a spiral staircase and things started fucking up because there were overlapping areas, areas that were too small, etc. For my particular case i could either enlarge the staircase or hide whatever is going on inside and make it a way simpler path
>tracking down desync issue between windows and linux clients >build a simulation validation system that checks whether the simulation doesn't deviate because of some hidden variable, but no luck, I'm not doing anything wrong there >build a replay system so I can replay the same gameplay on windows and linux machine >it always deviates at the exact point, spam debug prints outta the ass to find out where exactly >sqrt returns a different value for the two machines >the input is 0x1.fffffffffffffp-1, which is to say, the largest possible double precision float value that isn't 1.0. (aka 1 - 2^-53) >linux outputs the same value as input, windows outputs 1.0 >read documentation of sqrt() >std::sqrt is required by the IEEE standard to be exact. The only other operations required to be exact are the arithmetic operators and the function std::fma. After rounding to the return type (using default rounding mode), the result of std::sqrt is indistinguishable from the infinitely precise result. In other words, the error is less than 0.5 ulp. Other functions, including std::pow, are not so constrained. >required to be exact Am I getting fucking meme'd on?
>>1186118 # Is Linux really in the wrong here though? If I plug "(sqrt(1 - 2^-53) - (1 - 2^-53)) >= 2^-54" into wolfram, it returns false. I still have to read deeper into the standards on rounding logic, but I think that means its closer to the input value than the output value, which with a rounding mode of "FE_TONEAREST" (the default, which I verify before doing the math), should return just that. Meaning Windows is wrong. I should probably look what instructions it compiles to, and if it's just library calls, look into whatever the Windows function is doing. Maybe it has some of that x87 legacy in there for some stupid reason.
>>1186118 # >>1189449 # >Step through the assembly >Linux calls sqrtsd, the double-precision sse instruction, all good >Windows calls fsqrt, the 8087 instruction w h y windows, why must you embrace your legacy endlessly? I guess this explains why a few people desynced during demo day.
>>1200467 # Don't make me come over there you little shit. I'll have you know I graduated top of my class in the US navy seals, and I have over 300 confirmed shitposts.
>>1209745 # relaxing since my company just revealed vidya at gamescom and we're finally getting time off
>>1209745 # Im trying to work in a simple "push volume" in unreal 4 and its pissing me off, all I want it to do is "suck" you though a waterway when a fan is turned on, im sure in UDK you could just drag out a push volume with no need to set it up