00:29 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751207153246797895/92179734-33575a80-ee3d-11ea-82b4-2a5f2efe430c.png 00:29 <+bridge> [ddnet] i'll port this to ddn rn 05:20 <+bridge> [ddnet] <ᶰ°Konͧsti> Github costing monies now :justatest: 10:56 <+pastaaa> ist Kevin da? 10:59 <+Learath2> I'm sure there is atleast one Kevin around 11:50 <+bridge> [ddnet] > Github costing monies now :justatest: 11:50 <+bridge> [ddnet] @ᶰ°Konͧsti whut? 12:00 <+ChillerDragon> yea wdym @ᶰ°Konͧsti what costs? 12:01 <+bridge> [ddnet] chillerdragon someone on fng keeps faking u 12:41 <+bridge> [ddnet] @Learath2 could you please check https://github.com/ddnet/ddnet/pull/2675? 12:44 <+bridge> [ddnet] @deen it takes the 2k we use for binds and quadruples it 12:44 <+bridge> [ddnet] Actually 16k -> 64k 12:46 <+bridge> [ddnet] If we are okay with that for a seldom used feature, it just needs testing 12:47 <+bridge> [ddnet] yeah, seems like a bit much 12:47 <+bridge> [ddnet] but at least we should tell the guy how he can proceed 12:47 <+bridge> [ddnet] Problem is I'm not sure, I'd do some research at that point 12:50 <+bridge> [ddnet] You are better at this theoretical stuff than me. I would probably consider a flat vector of all binds we can linearly search through, a way to sort it so we can do binary search 12:51 <+bridge> [ddnet] Or since the last key always triggers the bind, maybe an array that we can index into which contains a list/vector of binds that end with this key? 12:51 <+bridge> [ddnet] Sounds complicated. Does it have such a big performance impact? 12:51 <+bridge> [ddnet] How about having all binds in a vector and just checking them one by one? 😄 12:52 <+bridge> [ddnet] Yeah that's the first one I proposed 12:53 <+bridge> [ddnet] sounds more complicated and less performant than the 64KB 12:53 <+bridge> [ddnet] I honestly don't know it's impact without testing but it's something that's definitely on the hot path, not sure if an O(n) worst case lookup is acceptable for us there 12:53 <+bridge> [ddnet] also average case 12:53 <+bridge> [ddnet] (but our n is fairly low, maybe we don't even notice it) 12:55 <+bridge> [ddnet] Eh, maybe the O(1) lookup is worth the extra 48k and the feature would be nice to have 12:57 <+bridge> [ddnet] ok, then let's stay with the 64 KB 12:58 <+bridge> [ddnet] we load all skins into memory, that has a way larger impact I guess 12:58 <+bridge> [ddnet] or the map data, uncompressed 13:00 <+bridge> [ddnet] I'll give it a test 14:31 <+bridge> [ddnet] @Learath2 any opinion on getting https://github.com/ddnet/ddnet/pull/2720 in? I'd really like to release today because of some fixes 14:32 <+bridge> [ddnet] the last screenshot looks fine 14:35 <+bridge> [ddnet] Sorry, I'd already looked at that and liked it, forgot to merge it 15:47 <+bridge> [ddnet] @Pure_luck ^^ that should do it 15:47 <+bridge> [ddnet] oh hey 15:48 <+bridge> [ddnet] heyo 15:48 <+bridge> [ddnet] i woke up late 15:56 <+bridge> [ddnet] Zwelf: Help, how to specify different syntax for sqlite than for mysql? 15:57 <+bridge> [ddnet] it's "INSERT OR IGNORE" in sqlite, but "INSERT IGNORE" in mysql 16:01 <+bridge> [freenode] Currently I handled all cases by writing a function like `const char *GetInsertIgnore() const` and MySQL would return "INSERT OR IGNORE" and SQLite would return "INSERT IGNORE". 16:03 <+bridge> [ddnet] @deen how does your skin patch handle people joining? 16:04 <+bridge> [ddnet] @Learath2 loads the skin, try it out 😄 16:04 <+bridge> [ddnet] I see how it adds a task to start the download, but Find only gets called once per tee joining/already ingame 16:04 <+bridge> [ddnet] first they are shown with default 16:04 <+bridge> [ddnet] I think it gets called more often 16:06 <+bridge> [ddnet] It actually does, huh how inefficient 16:12 <+bridge> [ddnet] `m_aSkins` is a sorted array and we are doing a linear search on it, every tick, for every tee 😄 16:14 <+bridge> [ddnet] Please optimize 😄 16:21 <+bridge> [ddnet] I see why, matricks' tl is actually really quite weak 16:30 <+bridge> [ddnet] <> there is no unpause button 16:30 <+bridge> [ddnet] Join Game should work on current DDNet version 16:30 <+bridge> [ddnet] not on old servers 16:31 <+bridge> [ddnet] <> ah ok i am on kog right now 16:31 <+bridge> [ddnet] tell them to update 16:40 <+bridge> [ddnet] :poggers: 16:42 <+bridge> [ddnet] KoG is missing a lot of the extended functionality from newer ddnet 16:43 <+bridge> [ddnet] they dont even have the ddnet network character 16:43 <+bridge> [ddnet] :cammostripes: 16:44 <+bridge> [ddnet] yeah, that's the problem here 16:44 <+bridge> [ddnet] Do we have a version check in the client? (and do we want to use this?) 16:44 <+bridge> [ddnet] Or should we just not show the pause button when server sent no ddnet character 16:45 <+bridge> [ddnet] not show it i guess 16:47 <+bridge> [ddnet] <ᶰ°Konͧsti> KoG is based on DDNet 9 16:47 <+bridge> [ddnet] <ᶰ°Konͧsti> lmfao 16:48 <+bridge> [ddnet] <ᶰ°Konͧsti> because DDNet built a lot of trash like the new hammer 16:48 <+bridge> [ddnet] <ᶰ°Konͧsti> Would be the worst thing in gores 16:48 <+bridge> [ddnet] lol no 16:48 <+bridge> [ddnet] <ᶰ°Konͧsti> lol yes 16:48 <+bridge> [ddnet] it would be good for them 16:48 <+bridge> [ddnet] its deterministic 16:48 <+bridge> [ddnet] u remove randomness 16:48 <+bridge> [ddnet] also its not just additions u see like hammer 16:48 <+bridge> [ddnet] it has lot of background improvements 16:48 <+bridge> [ddnet] <ᶰ°Konͧsti> First if all its not random 16:48 <+bridge> [ddnet] the ddnet character helps with prediction 16:49 <+bridge> [ddnet] <ᶰ°Konͧsti> Second u dont want freezed people hammer u when u hammer them 16:49 <+bridge> [ddnet] <ᶰ°Konͧsti> Definitely not 16:49 <+bridge> [ddnet] <ᶰ°Konͧsti> In gores 16:49 <+bridge> [ddnet] lol 16:49 <+bridge> [ddnet] the player can choose that 16:49 <+bridge> [ddnet] by pressing or not pressing hammer 16:49 <+bridge> [ddnet] <ᶰ°Konͧsti> players are dumb 16:49 <+bridge> [ddnet] that is a retarded take 16:49 <+bridge> [ddnet] <ᶰ°Konͧsti> new hammer is retarded, gores people would flame qdash to the hell 16:49 <+bridge> [ddnet] i dont want to argue with someone who thinks old ht is better than new ht 16:50 <+bridge> [ddnet] its like antimaskers 16:50 <+bridge> [ddnet] Can we not? You don't like anything new Konsti, we already got that 16:50 <+bridge> [ddnet] <ᶰ°Konͧsti> im just Saying new hammer would be a mess for KoG 16:50 <+bridge> [ddnet] <ᶰ°Konͧsti> And if u deny that rly 16:50 <+bridge> [ddnet] <ᶰ°Konͧsti> im a antimasker :justatest: 16:50 <+bridge> [ddnet] :) 16:53 <+bridge> [ddnet] @Jupstar ✪ "MY GAME IS BROKEN. EVERYWHERE IS RAINBOW." sounds like someone has graphics problems again 16:54 <+bridge> [ddnet] but rainbow = broken assets 16:54 <+bridge> [ddnet] normally 16:54 <+bridge> [ddnet] Or missing data file 16:54 <+bridge> [ddnet] yeah thats what i mean xd 16:54 <+bridge> [ddnet] Yeah, he left again, not sure what the problem way 16:54 <+bridge> [ddnet] was* 16:55 <+bridge> [ddnet] maybe we can add a mini chat responder 16:55 <+bridge> [ddnet] if someone says game is broken 16:55 <+bridge> [ddnet] send discord link 16:55 <+bridge> [ddnet] xD 16:57 <+bridge> [ddnet] nice update :D 16:57 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751456006973816872/Untitled.png 16:58 <+bridge> [ddnet] @TearlessAbyss46 what changed? 16:58 <+bridge> [ddnet] you can just go trashmap and make a random map name 16:58 <+bridge> [ddnet] then it shows on steam 16:58 <+bridge> [ddnet] eh? 16:58 <+bridge> [ddnet] its the map name 16:59 <+bridge> [ddnet] Who added that there? 16:59 <+bridge> [ddnet] heinrich 16:59 <+bridge> [ddnet] I thought we only got the steam name thing 16:59 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751456479311036486/unknown.png 17:00 <+bridge> [ddnet] :greenthing: 17:00 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751456555039457441/unknown.png 17:00 <+bridge> [ddnet] Heh, rich presence is there too, nice 17:00 <+bridge> [ddnet] whats new at entities? 17:00 <+bridge> [ddnet] @Slayer some small fixes because they are rendered differently 17:01 <+bridge> [ddnet] thanks 17:01 <+bridge> [ddnet] @Learath2 it should say Playing -mapname- tbh 17:02 <+bridge> [ddnet] ill check it 17:02 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751457167571419197/Untitled.png 17:02 <+bridge> [ddnet] why was this changed D: 17:02 <+bridge> [ddnet] what do u mean 17:02 <+bridge> [ddnet] tries to connect dummy 17:02 <+bridge> [ddnet] accidental disconnect 17:03 <+bridge> [ddnet] I think they're big enough not to misclick 17:03 <+bridge> [ddnet] I mean if you misclick that are you sure DDNet is the right game for you? 😄 17:04 <+bridge> [ddnet] good aim for noobs, use binds if you are pro 17:04 <+bridge> [ddnet] And since it was on the left newcomers always clicked Connect Dummy because they could find no other way to restart game 17:05 <+bridge> [ddnet] ps: spectate button width seems broken 17:05 <+bridge> [ddnet] ps: spectate button width SEEMS broken 17:05 <+bridge> [ddnet] why? 17:06 <+bridge> [ddnet] at the right 3 large buttons, at the left 2 small buttons and 1 large 17:06 <+bridge> [ddnet] you are killing perfectionist in me 17:06 <+bridge> [ddnet] Zwelf: No way to get the failed INSERTS from logs now, right? 17:06 <+bridge> [ddnet] we need the space 17:07 <+bridge> [ddnet] @BannZay Hm, could make it large size if space permits and only small for 5:4? But I thought same size is better because then translators instantly notice that their text won't fit 17:07 <+bridge> [ddnet] @deen inserts shouldn't(tm) fail, no? 17:08 <+bridge> [ddnet] They should go to the sqlite db if they "failed" 17:08 <+bridge> [ddnet] @Learath2 the sqlite insert syntax was wrong and no one noticed for a week 17:08 <+bridge> [ddnet] in sqlite it's INSERT OR IGNORE, not INSERT IGNORE 17:09 <+bridge> [ddnet] hope DDNet will you use adaptive UI one day, but its another story 17:11 <+bridge> [ddnet] DDNet7 is testable? I was thinking about packaging it to Arch, but I notice that the standard DDNet Client can't see DDNet7's servers, and DDNet7 is "Server only" according to the message the repository's description. 17:12 <+bridge> [ddnet] `You have to be on the 'GER' server to load this savegame` :justatest: 17:12 <+bridge> [ddnet] Is DDNet7 testable? I was thinking about packaging it to Arch, but I notice that the standard DDNet Client can't see DDNet7's servers, and DDNet7 is "Server only" according to the message the repository's description. 17:13 <+bridge> [ddnet] @RafaelFF DDNet7 is not really maintained anymore I think. Regular DDNet now supports 0.7 clients 17:13 <+bridge> [ddnet] so we don't run it anymore, had no changes in DDNet7 either 17:14 <+bridge> [ddnet] @Learath2 fast map download in 0.7 works well, thanks 17:14 <+bridge> [ddnet] that actually annoyed me today when testing somethintg 17:14 <+bridge> [ddnet] can you port savegames outside of the game somehow 17:15 <+bridge> [ddnet] saves are saved serverside obviously 17:15 <+bridge> [ddnet] otherwise u could edit them and put urself at finish 17:15 <+bridge> [ddnet] @NeXus to run them on your own server or why? 17:15 <+bridge> [ddnet] there is no ger server anymore 17:15 <+bridge> [ddnet] GER6 replaced GER 17:15 <+bridge> [ddnet] so the saves work there 17:15 <+bridge> [ddnet] ah good to know 17:15 <+bridge> [ddnet] let me actually rename it from GER6 to GER1 so that this isn't so confusing 17:16 <+bridge> [ddnet] noo 17:16 <+bridge> [ddnet] we could reach GER10 17:16 <+bridge> [ddnet] or rename it in the load message 17:16 <+bridge> [ddnet] @deen Download page still points to DDNet7. Should it be removed? 17:16 <+bridge> [ddnet] done 17:16 <+bridge> [ddnet] @NeXus also, empty servers don't require server name to match anymore, so you can load anywhere 17:17 <+bridge> [ddnet] @Ryozuki it could make sense if they are big and replication is expensive. CRC(or other idk) check should be enough if so 17:19 <+bridge> [ddnet] nononononono bad bad idea to rely on crc for that 😄 17:20 <+bridge> [ddnet] A cryptographic hash could be acceptable, but what's the point anyway 17:20 <+bridge> [ddnet] rather than that i would first change how saves are serialized, relying on %d\\t x100 is rly ugly 17:20 <+bridge> [ddnet] I guess it could be helpful while testing 17:21 <+bridge> [ddnet] what the size of such saves ? 17:21 <+bridge> [ddnet] https://github.com/ddnet/ddnet/blob/master/src/game/server/save.cpp#L226 17:21 <+bridge> [ddnet] i dont think size matters much also text can be compressed easily 17:22 <+bridge> [ddnet] We don't compress them though 17:22 <+bridge> [ddnet] I wonder why the saves aren't json yet tho 17:22 <+bridge> [ddnet] why json 17:22 <+bridge> [ddnet] It'd be so much easier to manipulate 17:23 <+bridge> [ddnet] It parses rather quick aswell 17:23 <+bridge> [ddnet] Better than `%d\t` 😄 17:23 <+bridge> [ddnet] why you want a save be easily manipulated 17:23 <+bridge> [ddnet] i would rather use a format that uses less space 17:23 <+bridge> [ddnet] binary tho 17:24 <+bridge> [ddnet] and is more machine friendly 17:24 <+bridge> [ddnet] zipped binary 17:24 <+bridge> [ddnet] to me its weird to store them as string 17:24 <+bridge> [ddnet] https://en.wikipedia.org/wiki/Truncated_binary_encoding 17:26 <+bridge> [ddnet] @Ryozuki Not like we are tight on space though, might aswell have it easily readable 17:26 <+bridge> [ddnet] why 17:26 <+bridge> [ddnet] only admins see the save 17:26 <+bridge> [ddnet] and u never modify them 17:26 <+bridge> [ddnet] We occasionally need to manipulate them, it'd atleast make that easier 17:26 <+bridge> [ddnet] ??? 17:26 <+bridge> [ddnet] are u saying 17:26 <+bridge> [ddnet] you cheated 17:26 <+bridge> [ddnet] ?? 17:27 <+bridge> [ddnet] 😄 17:27 <+bridge> [ddnet] :justatest: 17:27 <+bridge> [ddnet] :justatest: 17:27 <+bridge> [ddnet] https://github.com/servo/bincode 17:27 <+bridge> [ddnet] make a rust program 17:27 <+bridge> [ddnet] A couple occasions of stolen saves, a couple occasions of servers crashing as soon as the save is loaded 17:27 <+bridge> [ddnet] to manipulate them 17:27 <+bridge> [ddnet] what if you want to tell someone what their save is 17:27 <+bridge> [ddnet] :) 17:28 <+bridge> [ddnet] is not it better to make a tool to convert saved save into json instead of wasting disk space for readability? 17:28 <+bridge> [ddnet] A binary format would also make searching for someones saves painful as shit 17:28 <+bridge> [ddnet] I don't get what is your obsession with disk space, it's like 10-20 extra bytes per save 17:29 <+bridge> [ddnet] ok just use csv 17:29 <+bridge> [ddnet] why should we care 17:29 <+bridge> [ddnet] about anything 17:29 <+bridge> [ddnet] xd 17:32 <+bridge> [ddnet] well the metadata could stay in database for search and so, but I can not see much usecases for whole data being "readable" all the time 17:35 <+bridge> [ddnet] csv would work too 17:35 <+bridge> [ddnet] @BannZay metadata being extracted from the saves is even more unnecessary work 17:36 <+bridge> [ddnet] and it's duplication of data, another thing to go out of sync with the actual save 17:37 <+bridge> [ddnet] it's a matter of implementation 17:42 <+bridge> [ddnet] it's a matter of implementation (you can store only data not stored in db fields as binary. I am not familiar with current implementation) 17:44 <+bridge> [ddnet] @deen is the abuse of changing map names on a test server gonna be fixed? 17:45 <+bridge> [ddnet] what abuse? 17:45 <+bridge> [ddnet] Everyone can run their own server, Steam will show the map name you choose for your server 17:46 <+bridge> [ddnet] Can steam ban for this? Should not we show map name for players being on oficial DDNet servers? 17:47 <+bridge> [ddnet] like ban for setting status "BannZay playing on asshole" or so 17:47 <+bridge> [ddnet] @Jupstar ✪ uhm, you seem to have broken something with gl3.3 :/ 17:47 <+bridge> [ddnet] dejavu xD 17:47 <+bridge> [ddnet] on macOS atleast 17:48 <+bridge> [ddnet] non 3.3 works or what? 17:48 <+bridge> [ddnet] non 3.3 works just fine 17:48 <+bridge> [ddnet] and 3.3 doesnt load or looks buggy or what? 17:48 <+bridge> [ddnet] abc 3.3 17:49 <+bridge> [ddnet] 3.3 uses the same implementation for tile rendering as the other version now 17:49 <+bridge> [ddnet] Overlays on 17:49 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751469095047790618/screenshot_2020-09-04_18-48-49.png 17:50 <+bridge> [ddnet] Overlay off 17:50 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751469126744014998/screenshot_2020-09-04_18-48-56.png 17:50 <+bridge> [ddnet] initial 17:50 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751469141889646632/screenshot_2020-09-04_18-49-03.png 17:50 <+bridge> [ddnet] yeah looks broken xd 17:50 <+bridge> [ddnet] yeah verybroken 😄 17:50 <+bridge> [ddnet] have u disabled multi texturing? 17:50 <+bridge> [ddnet] in graphic settings tab 17:50 <+bridge> [ddnet] it says diusable on macos xD 17:50 <+bridge> [ddnet] xd 17:50 <+bridge> [ddnet] Ofc... 17:51 <+bridge> [ddnet] I have more than a single brain cell you know 17:51 <+bridge> [ddnet] xD 17:51 <+bridge> [ddnet] :justatest: 17:51 <+bridge> [ddnet] 62 fps 17:51 <+bridge> [ddnet] rip 17:52 <+bridge> [ddnet] "it says diusable on macos xD" - can we detect os and just disable it and not show ? 17:52 <+bridge> [ddnet] it probs even is 17:52 <+bridge> [ddnet] disabled default xd 17:53 <+bridge> [ddnet] I mean hide it in settings and dont let to set 17:53 <+bridge> [ddnet] why? 17:53 <+bridge> [ddnet] users are dumb you know 17:53 <+bridge> [ddnet] there are also new MacOS 17:53 <+bridge> [ddnet] not all will be broken 17:53 <+bridge> [ddnet] Sooooooooo, how are we debugging this? 17:53 <+bridge> [ddnet] renderdoc 17:54 <+bridge> [ddnet] oh no macos 17:54 <+bridge> [ddnet] yeah, you know 17:54 <+bridge> [ddnet] macos.. xd 17:55 <+bridge> [ddnet] i hate working for gpus xd 17:56 <+bridge> [ddnet] oh, man. The same on windows 17:56 <+bridge> [ddnet] What were you going to look at in renderdoc? 17:56 <+bridge> [ddnet] oh 17:56 <+bridge> [ddnet] even windows? 17:56 <+bridge> [ddnet] yes 17:56 <+bridge> [ddnet] then maybe bcs of multi textures? 17:56 <+bridge> [ddnet] off 17:56 <+bridge> [ddnet] i gonna look into the code 17:56 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751470806940385460/unknown.png 17:56 <+bridge> [ddnet] yeah let me see, never disabled it tbh xd 17:57 <+bridge> [ddnet] lol 17:57 <+bridge> [ddnet] and changed alot of code to be downward compatible to gl 2.1 17:59 <+bridge> [ddnet] @Jupstar ✪ it fixable as before by enabling Multiple texture unit (for windows) 17:59 <+bridge> [ddnet] ok ty will look into it, cant be that hard to fix 18:02 <+bridge> [ddnet] ok found it 18:02 <+bridge> [ddnet] \o/ 18:03 <+bridge> [ddnet] @deen also when you leave a map on steam it still shows you are on that map 18:03 <+bridge> [ddnet] multiple texture units makes it work on macOS too btw 18:03 <+bridge> [ddnet] just as long as you are okay with 10fps 18:04 <+bridge> [ddnet] what are ur fps without 18:05 <+bridge> [ddnet] @heinrich5991 @TearlessAbyss46 found a bug with map info on steam ^ 18:05 <+bridge> [ddnet] actually the multiple texture units was a try to remove API calls, but appearently some drivers implement them as software rendering to allow more units 18:05 <+bridge> [ddnet] pretty stupid xd 18:09 <+bridge> [ddnet] @Jupstar ✪ I get 62, but I cap it at 62 or it stutters AND turns my computer into a toaster 18:10 <+bridge> [ddnet] @Learath2 nice, cold winters can come 18:12 <+bridge> [ddnet] @Jupstar ✪ that fixes it 18:12 <+bridge> [ddnet] nice ty for testing 18:19 <+bridge> [ddnet] i think the hotfix doesnt hurry bcs its gl 3.3 only 18:19 <+bridge> [ddnet] maybe wait for skin fethcer or stuff like that 18:20 <+bridge> [ddnet] Nah, skin fetcher should be a big release 18:20 <+bridge> [ddnet] isnt 14.6.1 big release? xd 18:20 <+bridge> [ddnet] its 1 more than 14.6 18:20 <+bridge> [ddnet] :star15disappointed: 18:21 <+bridge> [ddnet] :lol: 18:29 <+bridge> [ddnet] ok valgrinded didnt show anything, hope its fine now 18:39 <+bridge> [ddnet] tbh i think changing `str_copy` straight up might just works for everything. but i'm a coward. 18:42 <+bridge> [ddnet] well not actually DoEditBox doesn't use str_copy at all lol 18:50 <+bridge> [ddnet] why dont we save `cl_overlay_entities` btw? 18:50 <+bridge> [ddnet] bcs nobody who uses entities would turn it on probably xd 18:52 <+bridge> [ddnet] why dont we save `cl_overlay_entities` btw? 18:52 <+bridge> [ddnet] bcs nobody who uses entities would turn it of probably xd 18:52 <+bridge> [ddnet] why dont we save `cl_overlay_entities` btw? 18:52 <+bridge> [ddnet] bcs nobody who uses entities would turn it off probably xd 19:03 <+bridge> [ddnet] I find it strange as well but I added it to autoexec instead of fixing in code. no idea why 19:03 <+bridge> [ddnet] we should really remove that C90 requirement xD 19:03 <+bridge> [ddnet] https://github.com/ddnet/ddnet/pull/2157 19:06 <+bridge> [ddnet] :EeveeShy: 19:06 <+bridge> [ddnet] > bcs nobody who uses entities would turn it off probably xd 19:06 <+bridge> [ddnet] @Jupstar ✪ but how have you tuned it on, lol ? 19:06 <+bridge> [ddnet] @deen didn't you release 14.6.1? 19:06 <+bridge> [ddnet] I got yelled at by C90 just now 19:06 <+bridge> [ddnet] > bcs nobody who uses entities would turn it off probably xd 19:06 <+bridge> [ddnet] @Jupstar ✪ but how have you tuned it on, lol ? Can we do it, please, can we save them? 19:06 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751488463039365260/unknown.png 19:07 <+bridge> [ddnet] https://github.com/ddnet/ddnet/blob/92616f1a55acc9c1cc1b63cc171dda6bed2781fb/src/engine/shared/config_variables.h#L296 19:07 <+bridge> [ddnet] add `|CFGFLAG_SAVE` 19:07 <+bridge> [ddnet] @BannZay if you convince the ppl that were holding it back yet, it'd be easy fix 19:08 <+bridge> [ddnet] but probs someone said its bad, and now they wont add it xd 19:08 <+bridge> [ddnet] I think there was another setting that is saved if you are using `cl_overlay_entites 100` 19:09 <+bridge> [ddnet] mh, I guess that died with ddrace client 19:10 <+bridge> [ddnet] i dont use ddnet long enough probs to know 😄 19:11 <+bridge> [ddnet] `MACRO_CONFIG_INT(ClShowEntities, cl_show_entities, 0, 0, 1, CFGFLAG_CLIENT, "Cheat to show game tiles")` 19:11 <+bridge> [ddnet] XDD 19:11 <+bridge> [ddnet] bcs it was a cheat xD 19:12 <+bridge> [ddnet] I am about to hate people saying its a cheat to such things 19:12 <+bridge> [ddnet] in fact ANY improvement is a cheat 19:12 <+bridge> [ddnet] This was alll the way back in ddrace times 😄 19:12 <+bridge> [ddnet] it was 10 years ago, that commit 19:17 <+bridge> [ddnet] just finished a map 19:17 <+bridge> [ddnet] . /teamrank 19:17 <+bridge> [ddnet] not showing 19:28 <+bridge> [ddnet] @Learath2 not yet, I went outside a bit 19:36 <+bridge> [ddnet] Hi 19:36 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751495854749122630/unknown.png 19:36 <+bridge> [ddnet] @mind that looks like a broken update :/ 19:37 <+bridge> [ddnet] That looks verybroken 19:37 <+bridge> [ddnet] we only removed the ca-ddnet.pem file *after* updating to the DDNet.exe that doesn't required 19:37 <+bridge> [ddnet] so I guess the DDNet.exe failed to get overwritten. Just download a new version I guess 19:37 <+bridge> [ddnet] Ok, thx for fast response 19:37 <+bridge> [ddnet] the autoupdater is kind of broken recently 19:37 <+bridge> [ddnet] not sure why 19:38 <+bridge> [ddnet] Next time we should make sure the autoupdater gets the ddnet folder exactly into the state we want 19:41 <+bridge> [ddnet] can we have overload for a str_format with return value char[] ? Many times we use buffer just for calling str_format 19:41 <+bridge> [ddnet] I mean is it possible in c++ ? 19:41 <+bridge> [ddnet] @BannZay system.c is written in C, not C++ 19:42 <+bridge> [ddnet] https://github.com/ddnet/ddnet/pull/2157 19:42 <+bridge> [ddnet] 19:42 <+bridge> [ddnet] the best closed pr 19:42 <+bridge> [ddnet] from deen 19:42 <+bridge> [ddnet] I don't think str_format can return a char[] even if it was C++, you can't return arrays in C++ either 19:43 <+bridge> [ddnet] how does char z[] = "123" work? The link would be enough 19:44 <+bridge> [ddnet] compiler knows the length of "123" 19:44 <+bridge> [ddnet] oh 19:44 <+bridge> [ddnet] then it'll just do char z[4] 19:44 <+bridge> [ddnet] probably can't use a variable length string other than manually calloc 19:44 <+bridge> [ddnet] but that way you'll need to free that afterwards. 19:52 <+bridge> [ddnet] What you are looking for would be something akin to `std::string` which sadly comes with a lot of heap allocation 19:55 <+bridge> [ddnet] stack is fast but I am worried about that amount of work processor does on each update. Like calling str_format N thouthands times per each Render tick 19:56 <+bridge> [ddnet] but thats bcs nothing in the client is not streamed xD 19:56 <+bridge> [ddnet] so much stuff could be buffered 20:05 <+bridge> [ddnet] I noticed something interesting. The way we have the borders actually inside the tiles means we always get the odd effect of tile borders being doubled up on the inside edges 20:06 <+bridge> [ddnet] How did mappers design around this for so long? I never really noticed it in almost any map 20:08 <+bridge> [ddnet] there is this borderfix tool 20:08 <+bridge> [ddnet] and alot of maps actually use vanilla resources 20:09 <+bridge> [ddnet] How does borderfix handle it? 20:11 <+bridge> [ddnet] this guy in #general 20:11 <+bridge> [ddnet] says he joins a server 20:11 <+bridge> [ddnet] and then the game crashes 20:12 <+bridge> [ddnet] any idea? 20:12 <+bridge> [ddnet] @Learath2 it will shrink down your image and than add a 2px border around each tile 20:12 <+bridge> [ddnet] How is that even supposed to help with the issue I mentioned 😄 20:13 <+bridge> [ddnet] @Ryozuki not really he can test gfx_opengl_major 1 20:13 <+bridge> [ddnet] but has to restart 20:14 <+bridge> [ddnet] before he joins xD else config not saved 20:14 <+bridge> [ddnet] 2011 pc 20:14 <+bridge> [ddnet] he says 20:14 <+bridge> [ddnet] :justatest: 20:14 <+bridge> [ddnet] @Jupstar ✪ I mean this 20:14 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751505601875673108/unknown.png 20:15 <+bridge> [ddnet] Does borderfix really help with that? I don't think so 20:15 <+bridge> [ddnet] not for squares.. except its exactly 2 pxs(the border) 20:16 <+bridge> [ddnet] @Jupstar ✪ doesnt work 20:17 <+bridge> [ddnet] imagine it like this: 20:17 <+bridge> [ddnet] u have bad resolution and the old renderer calculates the texel offset for a higher mipmap(and even wrongly)(but bad resolution = small mipmap) 20:17 <+bridge> [ddnet] => mipmap has to be fixed, by fixing the original image, else inconsistent in the border 20:17 <+bridge> [ddnet] @Ryozuki then he has a really broken GPU or its not realted to anything GPU side, maybe update GPU driver 20:17 <+bridge> [ddnet] if opengl 1.x fails its probs not tw 20:17 <+bridge> [ddnet] @Ryozuki I guess we need to see logs then 20:17 <+bridge> [ddnet] If you can get him to create logs. I find that rather difficult 20:18 <+bridge> [ddnet] @Jupstar ✪ he says it worked fine yesterday 20:18 <+bridge> [ddnet] maybe this update broke it for him 20:19 <+bridge> [ddnet] maybe he can try an older version again 20:19 <+bridge> [ddnet] thats strange, but the GL 1.x really doesnt use anything new 20:19 <+bridge> [ddnet] it even removes stuff that could break 20:19 <+bridge> [ddnet] @Learath2 appearently he already tried 1.x 20:19 <+bridge> [ddnet] Maybe it's not the gfx change but another change we have in 14.6? 20:19 <+bridge> [ddnet] yeah maybe 20:20 <+bridge> [ddnet] its weird 20:20 <+bridge> [ddnet] cuz it crashes when he joins a server 20:20 <+bridge> [ddnet] not in menus 20:20 <+bridge> [ddnet] it might have to do with the new tile rendering tbh 20:20 <+bridge> [ddnet] maybe the skin finder? 20:20 <+bridge> [ddnet] oh maybe 20:20 <+bridge> [ddnet] is it included? 20:20 <+bridge> [ddnet] no new tile renderer in GL 1.x 20:20 <+bridge> [ddnet] 100% the old one 20:21 <+bridge> [ddnet] Nah, the skin fetcher didn't make it yet 20:21 <+bridge> [ddnet] just the linear search is gone 20:21 <+bridge> [ddnet] Oh did that make it in? :d 20:21 <+bridge> [ddnet] i think in 16.5.1 it is for sure, isnt it? 20:21 <+bridge> [ddnet] It works just fine for me though and I don't see it crashing anything either 20:21 <+bridge> [ddnet] 14* 20:21 <+bridge> [ddnet] yeah i dunno, but that'd be something that is new and ingame 😄 20:22 <+bridge> [ddnet] @Jupstar ✪ btw, what happens now that we have correct rendering for the entire tile on all renderers? 20:22 <+bridge> [ddnet] now ravie and other mappers are happy xxd 20:22 <+bridge> [ddnet] bcs u can now do pixel perfect mapping 20:22 <+bridge> [ddnet] ravie complained 20:22 <+bridge> [ddnet] (almost pixel perfect) 20:22 <+bridge> [ddnet] its blurry i htink 20:22 <+bridge> [ddnet] xdd 20:23 <+bridge> [ddnet] u rly cant fix the blurryness? 20:23 <+bridge> [ddnet] its just blurry compared to old gl 3.3 20:23 <+bridge> [ddnet] not compared old renderer is it? 20:23 <+bridge> [ddnet] and all the maps with borderfixed tilesets will have a little thicker borders? 20:23 <+bridge> [ddnet] as said, you can fix the texture LOD in the shader if u want xd 20:23 <+bridge> [ddnet] u expect to much 20:23 <+bridge> [ddnet] from users 20:23 <+bridge> [ddnet] xd 20:23 <+bridge> [ddnet] @Learath2 yes, but all that use vanilla tile sets can be fixed 20:24 <+bridge> [ddnet] Hm, what causes the blurriness anyway? 20:24 <+bridge> [ddnet] by fixing the resource in the data 20:24 <+bridge> [ddnet] @Learath2 in OpenGL spec the LOD(mipmap level) is well defined, but drivers are free to choose the mipmap(most probs use some 3D algortihm, that makes in blurrier for 2D) 20:24 <+bridge> [ddnet] and the old GL 3.3 used the one from opengl 3.3 spec 20:25 <+bridge> [ddnet] the advantage of not choosing it manually is saving API calls and modifying state changes 20:25 <+bridge> [ddnet] How can LOD be well defined but not well defined at the same time? wtf 😄 20:25 <+bridge> [ddnet] and since 2D arrays dont require the LOD, it didnt do it 20:26 <+bridge> [ddnet] @Learath2 bcs in 3d u wont notice 😄 20:26 <+bridge> [ddnet] its better to use an efficient algorithm 20:26 <+bridge> [ddnet] bcs sampling is ofc expensive 20:27 <+bridge> [ddnet] How much of an overhead is choosing it manually? 20:27 <+bridge> [ddnet] not alot for teeworlds 20:27 <+bridge> [ddnet] in a 3d game it'd be more for sure 20:28 <+bridge> [ddnet] ok 20:28 <+bridge> [ddnet] it works 20:28 <+bridge> [ddnet] the previous version 20:28 <+bridge> [ddnet] but not this one 20:28 <+bridge> [ddnet] amazing 20:28 <+bridge> [ddnet] and the crash is when joining a server 20:28 <+bridge> [ddnet] I wonder what broke it 20:28 <+bridge> [ddnet] @Ryozuki are u sure he restarted after using 1.x? 20:29 <+bridge> [ddnet] i made sure 2 times 20:29 <+bridge> [ddnet] if he didnt lie 2 times then ye 20:29 <+bridge> [ddnet] the context isnt created on fly xD 20:29 <+bridge> [ddnet] ok 20:29 <+bridge> [ddnet] then its not the renderer i guess(except some buffer overflow valgrind didnt find) 20:29 <+bridge> [ddnet] what is crashing? how to repro? 20:29 <+bridge> [ddnet] @Ryozuki which OS is he on? 20:29 <+bridge> [ddnet] @deen I think it's just, join game, crash 20:30 <+bridge> [ddnet] windows 10 64bit 20:30 <+bridge> [ddnet] pc from 2011 20:30 <+bridge> [ddnet] im asking what cpu and gpu 20:31 <+bridge> [ddnet] @Jupstar ✪ I'm not really very familiar with how we do the rendering nowadays, do we even need/generate mipmaps? Isn't there usually a max LOD which we should be at when we are all the way up close? 20:31 <+bridge> [ddnet] @Learath2 20:31 <+bridge> [ddnet] https://www.khronos.org/registry/OpenGL/specs/gl/glspec33.core.pdf 20:31 <+bridge> [ddnet] 3.8.11 Texture Minification 20:31 <+bridge> [ddnet] i think that was the formula 20:31 <+bridge> [ddnet] Amd fx 6100 y nvidia geforse 7025 / a630 20:31 <+bridge> [ddnet] @Ryozuki it'd be nice if we can get the log 20:31 <+bridge> [ddnet] damn 20:31 <+bridge> [ddnet] @Learath2 wait gimme a minute and u'll laugh xd 20:33 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751510260014907412/screenshot_2020-09-04_20-33-24.png 20:33 <+bridge> [ddnet] this is not a bug 20:33 <+bridge> [ddnet] :lol: 20:33 <+bridge> [ddnet] its bcs without mipmaps zooming out far enough samples the texture so precice that it looks likek a bigger texture 20:33 <+bridge> [ddnet] @Ryozuki version from steam? 20:34 <+bridge> [ddnet] Lol wtf? 😄 20:34 <+bridge> [ddnet] @deen latest probs, i told him to download 14.5.1 and it worked 20:34 <+bridge> [ddnet] u cant execute the game without updating in steam so 20:34 <+bridge> [ddnet] Okay but I don't get why we aren't at max LOD when we are fairly close like this, there should be no scaling at all thus no blurriness? 20:35 <+bridge> [ddnet] the nice things about mipmaps is that it creates less aritfacts 20:35 <+bridge> [ddnet] linear interpolation is much worse than cubic 20:35 <+bridge> [ddnet] and mipmaps are probs cubic 20:35 <+bridge> [ddnet] Yeah mipmaps are fine, I just don't get why we get any blurriness at all up close 20:36 <+bridge> [ddnet] The resolution of the screen is probably closer to the first mipmap level 20:37 <+bridge> [ddnet] One could also use interpolation between the two mipmap levels. Might look sharper 20:37 <+bridge> [ddnet] That sounds extra unsharp 20:38 <+bridge> [ddnet] C♭ 20:39 <+bridge> [ddnet] Nope, usually looks better 20:39 <+bridge> [ddnet] How does opengl decide which mipmap level we are at? 20:39 <+bridge> [ddnet] I mean us being in level 0 when 10 tiles fill the screen isn't quite useful 20:42 <+bridge> [ddnet] @Jupstar ✪ 20:42 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751512488134180884/IMG_20200904_153002.png 20:42 <+bridge> [ddnet] he sees this 20:42 <+bridge> [ddnet] on 14.5.1 20:42 <+bridge> [ddnet] the weapon 20:42 <+bridge> [ddnet] xDDD 20:42 <+bridge> [ddnet] being a graphics programmer sure it tough 20:43 <+bridge> [ddnet] missing game.png? xD 20:44 <+bridge> [ddnet] If someone wants to test it: the tiles might look sharper if one replaces all occurrences of `GL_LINEAR_MIPMAP_NEAREST` with `GL_LINEAR_MIPMAP_LINEAR` in `backend_sdl.cpp` 20:45 <+bridge> [ddnet] or wrong resolution of game.png 20:46 <+bridge> [ddnet] How is interpolation supposed to look sharper than anything is beyond me... I hate what computer graphics has become 20:48 <+bridge> [ddnet] An API disconnected from hardware, drivers that implement the API so loosely it's impossible to get consistent results 20:48 <+bridge> [ddnet] Most screens have a lower resolution than the tiles, so OpenGL needs to scale it anyway. 20:49 <+bridge> [ddnet] Mipmaps just use the lower res images that were created with a better scaling algorithm 20:53 <+bridge> [ddnet] Gpus from different vendors implement certain thing in very different ways, so it's hard to design an API that works for all of them. With Vulkan there are even some issues because the API does not map very well to some mobile graphics chips 20:54 <+bridge> [ddnet] Actually at default zoom and for quite a bit atleast at 1080p we seem to be in LOD0 as I'd expect 20:58 <+bridge> [ddnet] its more about image quality 😄 20:58 <+bridge> [ddnet] At the zoom level `GL_LINEAR_MIPMAP_LINEAR` even would start to make a difference tiles are already too small for my eyes to detect the difference 20:59 <+bridge> [ddnet] yeah its hard to notice, tried it too, but depends really alot on resolution etc 20:59 <+bridge> [ddnet] I doubt resolution matters much the way you calculate the LOD 21:00 <+bridge> [ddnet] it does alot actually, bcs the screen of teeworlds is always same size 21:00 <+bridge> [ddnet] so less pixels are rendered 21:00 <+bridge> [ddnet] Even at the smallest resolution when tiles are waaay too small to begin with, it's firmly in LOD0 for like 6 zoom levels 21:00 <+bridge> [ddnet] what 21:00 <+bridge> [ddnet] i see the diff on 2k screen when zooming like 4 steps or smth already 21:01 <+bridge> [ddnet] it clearly chooses a different mipmap then 21:01 <+bridge> [ddnet] It could be the stupid retina screen messing with it 21:01 <+bridge> [ddnet] but didnt test with current version, let me give it a try 21:01 <+bridge> [ddnet] Or the fact that the window is actually small and not stretched in macOS 21:02 <+bridge> [ddnet] You can't have stretched fullscreen apps in macOS 21:03 <+bridge> [ddnet] already at 2nd zoom step 21:03 <+bridge> [ddnet] it uses different mipmap now 21:04 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751518029245972530/screenshot_2020-09-04_21-03-40.png 21:04 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751518031674343515/screenshot_2020-09-04_21-03-41.png 21:04 <+bridge> [ddnet] the kill tiles suddenly look shit af 21:04 <+bridge> [ddnet] but didnt profile to check for correctness, just feeling rn 21:06 <+bridge> [ddnet] @Ryozuki so what did i miss 21:06 <+bridge> [ddnet] is that the old or new version 21:06 <+bridge> [ddnet] old 21:06 <+bridge> [ddnet] well if even that is buggy 21:06 <+bridge> [ddnet] new crashes 21:06 <+bridge> [ddnet] he really needs to update his driver 21:06 <+bridge> [ddnet] i'm not going to add software renderer xDDDDD 21:06 <+bridge> [ddnet] u cant expect to play games with a potato 21:06 <+bridge> [ddnet] from last century 21:07 <+bridge> [ddnet] i mean GL 1.x is litterally from last century xD 21:07 <+bridge> [ddnet] 1.2.1 14. Oct. 1998 21:07 <+bridge> [ddnet] how hard it is to port this game on android ? 21:08 <+bridge> [ddnet] to be able to spectate 21:08 <+bridge> [ddnet] mhh, you need to add OpenGL es support, that isnt too hard probs 21:08 <+bridge> [ddnet] and some libs 21:08 <+bridge> [ddnet] but generally its not hardest, just annoying 21:09 <+bridge> [ddnet] if you have a modern phone, the opengl 3.3 support should even work with android 21:09 <+bridge> [ddnet] opengl es 3.0 21:09 <+bridge> [ddnet] The draw call count might cause issues. Mobile chips often can't really handle more than 100 per frame. Today the game easily exceeds 200 21:10 <+bridge> [ddnet] modern phone are better than that above guys PC XD 21:10 <+bridge> [ddnet] Meh, I guess I'm not qualified to comment on this with my current gl knowledge. Tilesets looking blurry at the default zoom is just confusing to me 21:11 <+bridge> [ddnet] and IMHO shouldn't happen 21:11 <+bridge> [ddnet] Still they have some limits that dedicated gpus don't have. Modern engines just handle things differently 21:11 <+bridge> [ddnet] nice @mind 21:11 <+bridge> [ddnet] hoped for that one 😄 21:11 <+bridge> [ddnet] yes, but really modern phones aren't really bad anymore 21:12 <+bridge> [ddnet] True 21:12 <+bridge> [ddnet] Are we planning to port SDL2 DDNet to android again? 😄 21:12 <+bridge> [ddnet] Would be nice to have a watch + chat mode 21:12 <+bridge> [ddnet] if we wouldnt need to fix so much gui stuff, we could do it 21:12 <+bridge> [ddnet] but the code base gets really ugly xD 21:13 <+bridge> [ddnet] i bet my 5 years old mobile phone can handle with gl 3.3 with atleast 30fps xD 21:14 <+bridge> [ddnet] (or better GL 3.0 es, basically the same) 21:15 <+bridge> [ddnet] Guess I'll have to learn gl myself 21:16 <+bridge> [ddnet] GL is just about vertices xd 21:16 <+bridge> [ddnet] u have to feel it 21:16 <+bridge> [ddnet] look around your room and see vertices 21:16 <+bridge> [ddnet] @Jupstar ✪ just get rid of the gui? Only need server browser and then ingame 21:16 <+bridge> [ddnet] @deen sure would be nice too 21:18 <+bridge> [ddnet] does somebody have the SVG for the original tilesets btw? 21:18 <+bridge> [ddnet] would be cool to test 4k versions 21:18 <+bridge> [ddnet] Not public 21:18 <+bridge> [ddnet] Ask in tw discord, I think zatline has them 21:18 <+bridge> [ddnet] Meh, that's sad to hear for an open source game 21:19 <+bridge> [ddnet] Let's commission replacements? 21:20 <+bridge> [ddnet] i mean why would they care 21:20 <+bridge> [ddnet] recreating them in inkscape isnt really hard 21:20 <+bridge> [ddnet] just annoying work xd 21:25 <+bridge> [ddnet] Android version could evolve into something bigger then just spectator. Some simple maps might be finished using android inputs. There is only 4 buttons and aim thing to finish map. Check out something like brawl stars. 21:26 <+bridge> [ddnet] but is it fun? xd 21:27 <+bridge> [ddnet] if it is not so laggy I believe so 21:28 <+bridge> [ddnet] Attach it to a screen and play with Bluetooth keyboard or a controller 😅 21:29 <+bridge> [ddnet] I cannot do it being at work :( 21:31 <+bridge> [ddnet] Or outside... you got the idea 21:44 <+bridge> [ddnet] bug, after apply "cl_text_entities_size 40" 21:44 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751528087107272764/unknown.png 21:44 <+bridge> [ddnet] @Matodor should be fixed in 14.6.1. are you on that. Otherwise @Jupstar ✪ ^ 21:47 <+bridge> [ddnet] I think instead of Kill and Pause the new buttons should say something more descriptive like Restart and Watch 21:49 <+bridge> [ddnet] yes should be fixed 21:50 <+bridge> [ddnet] thx 😎 22:07 <+bridge> [ddnet] @Ryozuki btw i just read the msg again, 14.5.1, so that means he has a opengl 1.x card, bcs 2.x card normally support mipmapping, and since his fails its probs just that it should fallback in the current version. 22:07 <+bridge> [ddnet] 22:07 <+bridge> [ddnet] So gfx_opengl_major 1 => restart should actually fix it WTF, 22:07 <+bridge> [ddnet] and learaths binary search isnt in yet, so he was 2 times lieing probably xD 22:11 <+bridge> [ddnet] xD 22:11 <+bridge> [ddnet] can he speak english? 22:15 <+bridge> [ddnet] idk 22:52 <+bridge> [ddnet] its a wasted effort 22:53 <+bridge> [ddnet] idc, i want it fixed for everyone 22:53 <+bridge> [ddnet] else they cant play teeworlds 22:53 <+bridge> [ddnet] :tear: 23:33 <+bridge> [ddnet] just found out the proof circle in editor is off by quite a bit 23:33 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751555506614829176/proof.png 23:39 <+bridge> [ddnet] hm, that looks like it should be totally centered in the editor.cpp 23:42 <+bridge> [ddnet] And DrawCircle looks correct too 23:42 <+bridge> [ddnet] maybe we don't put the tee in the middle? 23:45 <+bridge> [ddnet] I mean we could just manually move the circle up a bit until it about fits, but that's still weird 23:46 <+bridge> [ddnet] it looks centered on the red squares 23:47 <+bridge> [ddnet] @deen the "circle" part of your tee is not exactly centered, it's a bit higher 23:47 <+bridge> [ddnet] @Ravie could you send me the map? 23:47 <+bridge> [ddnet] I'm horrible at mapping 😄 23:48 <+bridge> [freenode] > Zwelf: No way to get the failed INSERTS from logs now, right? 23:48 <+bridge> [freenode] no not really, you can grep for "save score failed for all servers" and "save team score failed for all servers" to count all lost ranks, and the finish time should be printed some lines above those. 23:48 <+bridge> [ddnet] Yeah, counted > 200 since yesterday, so no chance 23:48 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/751559300056612874/proof_circle.map 23:48 <+bridge> [ddnet] It's also weird that sql connection fails for all servers, no matter where they're located 23:48 <+bridge> [ddnet] our network connection can't be that bad, can it? 23:48 <+bridge> [ddnet] maybe the tcp connection is missing a keepalive? 23:49 <+bridge> [freenode] Are some logs lost as well? 23:49 <+bridge> [ddnet] yes, we rotate logs every day 23:49 <+bridge> [ddnet] so no logs before then 23:50 <+bridge> [ddnet] So did we lose 200 ranks? :/ 23:51 <+bridge> [ddnet] no, > 1000 probably 23:51 <+bridge> [ddnet] we've been running this version for a while 23:51 <+bridge> [freenode] If the database connection fails due to no activity for some time, it tries to connect again. 23:52 <+bridge> [freenode] I guess, Only a teehistorian parser could bring those ranks back. 23:52 <+bridge> [ddnet] Well is it fixed now? 23:53 <+bridge> [ddnet] teehistorian sadly doesn't save finishes, so not just a parser but a way to actually play them back would be needed 23:53 <+bridge> [ddnet] on restarted servers it should be fixed 23:54 <+bridge> [ddnet] The connection failing is interesting too now that ddnet.tw pretty much is never overloaded/ddosd 23:55 <+bridge> [ddnet] Not like it's common that major data centers have no routes left between them either 23:56 <+bridge> [ddnet] uh, the new update has overwritten old entities which I spent 2 hours on customizing it 23:56 <+bridge> [ddnet] and I didnt back it up 23:56 <+bridge> [ddnet] am I fucked? 23:56 <+bridge> [ddnet] Eh, I think you are 23:56 <+bridge> [ddnet] rip murpi 23:56 <+bridge> [ddnet] hi sheep 23:57 <+bridge> [ddnet] :wowcry: 23:57 <+bridge> [ddnet] Maybe a NTFS file recovery thing can try to salvage it, but it's very very likely that it used the exact same physical disk space while updating 23:57 <+bridge> [ddnet] @murpi sorry :/ 23:57 <+bridge> [ddnet] Your custom files go inside the config directory :/ 23:57 <+bridge> [ddnet] We really should make that more clear 😦 23:58 <+bridge> [ddnet] Or we could ask people before overwriting their customized files? 23:58 <+bridge> [ddnet] If we have the hashes of each file, we can check while updating 23:58 <+bridge> [ddnet] We don't know that the files are customized, atleast not as it is right now 23:58 <+bridge> [ddnet] So let's store hashes of everything in data 23:58 <+bridge> [ddnet] This happened every time we updated entities 23:59 <+bridge> [ddnet] so we move from update6 to update7 and include an old_hash key for each file? 23:59 <+bridge> [ddnet] If old hash doesn't match we ask for a warning