00:01 <+bridge> [ddnet] @deen https://github.com/bors-ng/bors-ng/issues/185 didn't have movement for three years though 00:10 <+bridge> [ddnet] @deen oh no, the updater requires http.h, which is not in 0.7 00:10 <+bridge> [ddnet] so I guess I need that too :/? Uhm... I think this wont happen if no one PRs it :D 01:22 <+bridge> [ddnet] @Learath2 i need your help/opinion... I am in a conflict with myself regarding 128 players. I got it to a point where it works pretty good, also with dummy and so on. I made it so that players with the same IP get the same faked IDs in the map. Vanilla clients 0.7 works fine of course, and I made an adjustment in my 0.7 F-Client for some things for the dummy (because everything is different, e.g. team changes, client enter, drop etc. ever 01:23 <+bridge> [ddnet] DDNet clients work fine too, also with dummy. Expect one point: whisper. When whispering to the dummy that is not active at the moment, it will of course say the name of the player with the fake id of the current player's map. I could fix that by sending RECV whispers as server messages for DDNet clients. 01:24 <+bridge> [ddnet] But generally, I am not 100% sure whether I should keep the whole thing or go back to 64 players. I mean, it definitely is a nice thing to have, since for example today I had 80 players at a time. Usually tho, it is around 60-70 01:26 <+bridge> [ddnet] @Learath2 did you figure out why bors chokes on some PRs? 01:29 <+bridge> [ddnet] @heinrich5991 well it was choking on that one because it was modifying a file in .github/workflows, apparently integrations are not allowed to make changes to that 01:30 <+bridge> [ddnet] ah yea: https://github.community/t5/GitHub-Actions/GitHub-action-Resource-not-accessible-by-integration/m-p/19950/highlight/true#M75 01:30 <+bridge> [ddnet] @fokkonaut I'd stick to 64 players, the translation patches were never nice, I think 64 is more then enough 01:30 <+bridge> [ddnet] ty 01:30 <+bridge> [ddnet] @Learath2 well, BlmapChill is a big map tho, and 128 is working fine besides that 01:31 <+bridge> [ddnet] but i also think the algorithm is a bit laggy 01:31 <+bridge> [ddnet] this is all fixable but it'll take a lot of work, to get the algorithms all optimized 01:33 <+bridge> [ddnet] I don't get your whisper issue, the current players id map should match the dummies exactly 01:33 <+bridge> [ddnet] No, of course it doesnt on the serverside 01:34 <+bridge> [ddnet] if the dummy is in another area where other tees are 01:34 <+bridge> [ddnet] they have different id's on dummy/main 01:34 <+bridge> [ddnet] yeah but the first part is the same 01:34 <+bridge> [ddnet] its about WHO sent the message 01:34 <+bridge> [ddnet] oh, do we send a client id and not an actual name? 01:35 <+bridge> [ddnet] yes, thats how ddnet whisper works. 01:35 <+bridge> [ddnet] so as i said, i could fix it by sending a servermsg to all clients with the same IP as the receipiant 01:35 <+bridge> [ddnet] why all clients? you as the server always know the correct recipient 01:36 <+bridge> [ddnet] Yes, but if I send it to the dummy only, the main wont receive it 01:36 <+bridge> [ddnet] only whisper and team messages are received when they go to the udmmy 01:36 <+bridge> [ddnet] this all sounds very meh 01:38 <+bridge> [ddnet] yea. but okay, another thing. Lets imagine I remove 128 support. F-Client still has some things in the client, for example a complete set of m_aClients[] for thedummy, so that the dummy can have different infos for each client, that was of course used for 128 players, as the info is not in snaps anymore. But if I remove that, also team changes for specific tees to specific tees only wouldnt work in 0.7 dummy anymore. My mod does team chan 01:38 <+bridge> [ddnet] i am completely overwhelmed right now xD 01:39 <+bridge> [ddnet] i hate 0.7 for moving stuff out of snaps 01:41 <+bridge> [ddnet] with 64 clients you should never be lacking info 01:42 <+bridge> [ddnet] yes, but if my dummy joins a minigame, it gets teamchanges to team blue for everybody thats not in the same minigame, when swapping to the main again, the others are of course still in team blue, which is false. 01:42 <+bridge> [ddnet] The second m_aCleints array fixes that 01:42 <+bridge> [ddnet] but i inititally did that for 128 players, so i think i would need to keep that too in the client, even if i remove 128 01:42 <+bridge> [ddnet] having different sets information for dummy and main sounds logical to me 01:42 <+bridge> [ddnet] yea 01:43 <+bridge> [ddnet] This is a point where I really notice how much better the Snap thing is 01:43 <+bridge> [ddnet] @heinrich5991 why? the dummy and the main client exist in the same game world 01:43 <+bridge> [ddnet] But for cases like the team changes it is neccessary 01:44 <+bridge> [ddnet] @Learath2 in every server implementation, currently 01:44 <+bridge> [ddnet] but if I write my own tw server, there's no guarantee I send the same IDs to every player 01:45 <+bridge> [ddnet] e.g. I could choose to send cid=0 for the client's own tee 01:46 <+bridge> [ddnet] Well I don't like these kinds of "many mods in one mod" mods to begin with tbh 01:46 <+bridge> [ddnet] @heinrich5991 I don't think this is a good idea 01:46 <+bridge> [ddnet] why not? 01:46 <+bridge> [ddnet] adding 50 layers of translation is just not a good idea 01:46 <+bridge> [ddnet] @Learath2 well, teeworlds adventure mod also does this, they have multiworld and also send faked ids 01:47 <+bridge> [ddnet] then I don't like the "teeworlds adventure mod" either 01:47 <+bridge> [ddnet] XD 01:48 <+bridge> [ddnet] @heinrich5991 why would you want inconsistent ids anyway, it would only give you more players in one server, or support some mod in mod situations 01:48 <+bridge> [ddnet] because I don't want to leak server state e.g. 01:48 <+bridge> [ddnet] dummy is only a mod too, it should work as a real new client i think 01:49 <+bridge> [ddnet] no harm to leaking the actual client ids 01:50 <+bridge> [ddnet] only more complexity with designing the translation layer that'll just add overhead 01:51 <+bridge> [ddnet] "runtime overhead" meh 01:51 <+bridge> [ddnet] "runtime overhead"? meh 01:51 <+bridge> [ddnet] at the very best case I can imagine (a swap from your cid to 0 and put the guy at 0 in your position (at that point why bother you are still leaking state)) it's still a constant overhead of an extra branch instruction 01:52 <+bridge> [ddnet] you shouldn't worry about that extra instruction 01:52 <+bridge> [ddnet] you should worry about making a proper collision algorithm first 01:52 <+bridge> [ddnet] (that is not O(n^2). that costs more instructions) 01:52 <+bridge> [ddnet] At no gain, it's still an extra instruction 01:53 <+bridge> [ddnet] say you want to swap tees in 0.7 01:53 <+bridge> [ddnet] then you need your ID swapping feature anyway 01:53 <+bridge> [ddnet] why? 01:54 <+bridge> [ddnet] ah, you could copy over the tee data, fair enough 01:54 <+bridge> [ddnet] any useful translation table and you are looking at some very bad performance characteristics 01:54 <+bridge> [ddnet] why? 01:54 <+bridge> [ddnet] at the very least while generating the table 01:56 <+bridge> [ddnet] if you want to support more then 64 clients, you are going to have to sort the tees based on distance, that's O(logn) with a good algorithm 01:57 <+bridge> [ddnet] If you want just a 64 -> 64 mapping, it'll have a [64][64] memory cost 01:58 <+bridge> [ddnet] which is 4KB btw 01:58 <+bridge> [ddnet] besides, put aside the cost, we don't seem to care about that much in 2020, it doesn't bring anything to the table, I don't see a good use for it 01:58 <+bridge> [ddnet] but okay, I see that you don't need it, so it's unnecessary overhead 01:58 <+bridge> [ddnet] but I do think that you worry about the wrong costs if you're worried about one extra mispredicted branch per network packet 01:59 <+bridge> [ddnet] this is how these things add up, we have all become too lax, started relying on the processors raw power to take care of things, fancy branch predicting algorithms in cpus, speculative execution 02:00 <+bridge> [ddnet] I think you're setting the wrong targets 02:00 <+bridge> [ddnet] I'm sure I can find a branch like this in the handcoded assembly game rct2, for example 02:00 <+bridge> [ddnet] which is an example for very good performance 02:01 <+bridge> [ddnet] if they need the branch, then it's fine 02:01 <+bridge> [ddnet] I'm sure I'll find unneeded branches in there 02:01 <+bridge> [ddnet] e.g. reducing the number of syscalls, that's a lot better target 02:01 <+bridge> [ddnet] these cost actual time 02:01 <+bridge> [ddnet] especially with spectre/etc. fixes 02:02 <+bridge> [ddnet] Ofc they are more significant 02:03 <+bridge> [ddnet] this is all beside the point though, it's only a single branch insruction in the silly no-op translation "table" 02:03 <+bridge> [ddnet] yes, but I find it ridiculous that you claim that that branch matters 02:04 <+bridge> [ddnet] I agree that adding a translation table is not worth it for no benefit 02:04 <+bridge> [ddnet] but claiming that a branch there matters, is absurd IMO 02:04 <+bridge> [ddnet] depends on how you define "matters" 02:04 <+bridge> [ddnet] matters for software performance 02:04 <+bridge> [ddnet] for making your software run faster on a '95 computer 02:05 <+bridge> [ddnet] that branch instruction will cost you a couple cycles, and it will have no gain, I don't see how anyone can argue that that is fine 02:06 <+bridge> [ddnet] If there was any value gain, fine there is nothing to argue, the branch is insignificant 02:06 <+bridge> [ddnet] say the gain is that I want every client id to be 0 02:06 <+bridge> [ddnet] that is the only reason 02:06 <+bridge> [ddnet] even then it's completely insignificant 02:06 <+bridge> [ddnet] I can't state enough wishes of that form to make it matter 02:06 <+bridge> [ddnet] if you need that, sure branch away 02:07 <+bridge> [ddnet] it's just that I find cid=0 aesthetically more pleasing 02:07 <+bridge> [ddnet] even this very discussion probably used up more cycles than the branch in my hypothetical tw server 😛 02:07 <+bridge> [ddnet] by broadcasting it to many people, etc. etc. 02:08 <+bridge> [ddnet] so in a sense, arguing about that branch made performance worse already 02:09 <+bridge> [ddnet] I don't get the point of this, every decision you make comes with a cost and a reward, you try to maximize reward and minimize cost 02:09 <+bridge> [ddnet] except that you have a meta optimizer there, too 02:10 <+bridge> [ddnet] that makes you not care about details that don't matter 02:10 <+bridge> [ddnet] otherwise you'd spend too much time optimizing things that don't matter 02:10 <+bridge> [ddnet] as a matter of fact, this optimizer just called, sorry 02:13 <+bridge> [ddnet] Premature optimization is not a good idea, sure. Not being cognisant of the performance implications of your code is worse imho 02:16 <+bridge> [ddnet] in context of teeworlds we have the luxury of being lax with our code because we have many cycles to spare before the performance becomes an issue to players 02:17 <+bridge> [ddnet] can't apply the same attitude working with high performance software where you want to squeeze out as much performance as you can nor can you apply it to embedded programming where you don't have performance to throw out the window 02:20 <+bridge> [ddnet] branch mispredictions and cache misses have non zero cost, they don't matter much for something like teeworlds that lies in the middle of the spectrum, that doesn't mean we should shove a loop that branches a million times in the middle of the main loop 02:21 <+bridge> [ddnet] I agree with your previous messages 02:22 <+bridge> [ddnet] anyway this is a tangent, I don't think we should decouple the clients understanding of the gameworld from the servers 02:23 <+bridge> [ddnet] Having a seperate set of clients for dummy and main is just like having another client as dummy 02:24 <+bridge> [ddnet] it just makes sense imo (for 0.7, in 0.6 it is automatically due to snaps) 02:25 <+bridge> [ddnet] if you think clients shouldn't be receiving the same set of clients, yes, that is the correct approach 02:25 <+bridge> [ddnet] for mods you can never kniw 02:25 <+bridge> [ddnet] know* 02:28 <+bridge> [ddnet] I also still believe the player info belonged in the snap 02:28 <+bridge> [ddnet] Me too 02:30 <+bridge> [ddnet] apparently the new skin system made the snap a bit larger 02:30 <+bridge> [ddnet] Is it so much bigger? 02:31 <+bridge> [ddnet] idk, I never checked it 02:31 <+bridge> [ddnet] guess a few new entries is not too bad 02:31 <+bridge> [ddnet] still, delta compression should be dropping the entire playerinfo 02:31 <+bridge> [ddnet] wym? 02:32 <+bridge> [ddnet] we don't always send complete snapshots 02:32 <+bridge> [ddnet] most of them are diffs called deltashots 02:32 <+bridge> [ddnet] Why? 02:33 <+bridge> [ddnet] because it'd be a huge amount of traffic to keep sending entire snapshots? 02:33 <+bridge> [ddnet] Oh yea true 02:36 <+bridge> [ddnet] I'm very sleepy, I have an upset stomach so I can't sleep 02:36 <+bridge> [ddnet] so annoying 14:13 <+bridge> [ddnet] You need to enable the python language server in vscode to get proper intellisense... 14:28 <+bridge> [ddnet] ofc 14:34 <+bridge> [ddnet] not microsofts one, that one doesn't want to do type inference 14:34 <+bridge> [ddnet] you need to enable the jedi one 14:38 <+bridge> [ddnet] also protip: if any of you wants a windows 10 virtual machine, go for a windows 7 one instead it performs much smoother under virtualization 14:38 <+bridge> [ddnet] if you have to get a windows 10 one, boot the vm headless and RDP into the machine instead, microsofts rdp works better then all 3 virtualization platforms I tried 14:39 <+bridge> [ddnet] (their native display adapters that is) 14:40 <+bridge> [ddnet] can you rdp from linux? 14:42 <+bridge> [ddnet] hm, didn't test it from linux, macOS has an official client, so does iOS, I think linux has a couple unofficial alternatives that perform well but I can't vouch for those 14:51 <+bridge> [ddnet] Is there a tool better then bors? If not I'm thinking of doing https://github.com/bors-ng/bors-ng/issues/185 15:34 <+bridge> [ddnet] @Learath2 if you find a better tool, let me know 16:45 <+bridge> [ddnet] hm, I really expected more from pythons functional programming interfaces, you can't even easily chain 16:46 <+bridge> [ddnet] Java and javascript both handle this beautifully 16:48 <+bridge> [ddnet] `itertools.chain`? 16:49 <+bridge> [ddnet] chain as in chain higher order functions like filter, map, reduce 16:52 <+bridge> [ddnet] maybe I don't know what chain is 16:52 <+bridge> [ddnet] do you mean flatten? 16:52 <+bridge> [ddnet] `list.stream().map(m -> sin(m)).filter(m -> m > 0).collect(toList())` is what it'd look like in Java 16:53 <+bridge> [ddnet] `list.map(m => sin(m)).filter(m => m > 0)` in js 16:53 <+bridge> [ddnet] `[sin(m) for m in list if sin(m) > 0]` pythonic 16:53 <+bridge> [ddnet] `list(filter(lambda m: m > 0, list(map(lambda m: sin(m)))` in python 16:54 <+bridge> [ddnet] ofc I can use a list comprehension, my complaint was about the higher order functions filter, map and reduce 16:54 <+bridge> [ddnet] hm 16:55 <+bridge> [ddnet] I guess python's stance on this is: use comprehensions instead of higher order functions? 16:55 <+bridge> [ddnet] hm, if that is your stance about something in your language, I don't think it belongs in the language 16:56 <+bridge> [ddnet] list comprehensions have some problems 16:56 <+bridge> [ddnet] I was thinking that list comprehensions are equivalent in functionality 16:58 <+bridge> [ddnet] yes, lambdas are too verbose for other things in python 16:58 <+bridge> [ddnet] I'm pretty sure they are 17:01 <+bridge> [ddnet] well javascript lacks it but java and rust have lazy evaluation for their streams and iterables respectively 17:01 <+bridge> [ddnet] put it in paranthesis in python to achieve that 17:01 <+bridge> [ddnet] put what in parenthesis? 17:01 <+bridge> [ddnet] the list comprehension 17:02 <+bridge> [ddnet] (for x in a) 17:02 <+bridge> [ddnet] `(sin(a) for a in iterable)` 17:02 <+bridge> [ddnet] hm, hadn't seen that before, that's useful 17:03 <+bridge> [ddnet] this makes a generator iirc 17:03 <+bridge> [ddnet] yes 17:07 <+bridge> [ddnet] fwiw I prefer the chained calls to the list comprehension syntax, it's just easier on the eyes imho 17:07 <+bridge> [ddnet] I guess one could argue that pythons list comprehensions are "easier" to read as it's basically english 17:07 <+bridge> [ddnet] oh and list comprehensions are just awkward to type 17:10 <+bridge> [ddnet] Overall, I think if I wanted a language that lacks a type system, I think I'd use javascript nowadays 17:10 <+bridge> [ddnet] Next up on the docket of things to try is perl 17:11 <+bridge> [ddnet] js has its weird things but its good 17:11 <+bridge> [ddnet] i use typescript if i can tho 17:11 <+bridge> [ddnet] I also go for ts for any project that's getting deployed 17:11 <+bridge> [ddnet] also the way it seamlessly connects with html when using react, jsx tsx 17:11 <+bridge> [ddnet] i love it 17:15 <+bridge> [ddnet] observables and promises aren't half bad constructs for doing asynchronous work too 17:19 <+bridge> [ddnet] nim's futures are also cool 17:20 <+bridge> [ddnet] maybe I can take a look at Go, it's supposed to be great for concurrency 17:44 <+bridge> [ddnet] tee keep crying now when you hold fire while frozen 17:44 <+bridge> [ddnet] is that wanted behavior? 17:53 <+bridge> [ddnet] no 18:03 <+bridge> [ddnet] lol, how did that happen? 😄 18:03 <+bridge> [ddnet] the same commit as before, I guess 18:03 <+bridge> [ddnet] the same PR 18:04 <+bridge> [ddnet] idk what "as before" refers to in this context 18:04 <+bridge> [ddnet] also do you know of a python equivalent for an assignment in an if stmt? 18:04 <+bridge> [ddnet] https://github.com/ddnet/ddnet/pull/2086 18:04 <+bridge> [ddnet] or if let in rust 18:04 <+bridge> [ddnet] they implemented the "walrus operator" recently 18:05 <+bridge> [ddnet] if m := re.match("abc", "abc"): 18:05 <+bridge> [ddnet] it's >= 3.8 sadly 18:05 <+bridge> [ddnet] `if m := re.match("abc", "abc"):` 18:05 <+bridge> [ddnet] ah 18:05 <+bridge> [ddnet] too recent 18:05 <+bridge> [ddnet] other than that, not aware of any 18:05 <+bridge> [ddnet] is the idiom really `m = re.match() if m is not None:`? 18:06 <+bridge> [ddnet] yes 18:06 <+bridge> [ddnet] or `if m:` also works 18:06 <+bridge> [ddnet] this language was made by the scope leak gang 🙂 18:07 <+bridge> [ddnet] there is only¹ function scope in python anyway 18:07 <+bridge> [ddnet] yeah, needs more scoping 18:07 <+bridge> [ddnet] scopes are nice to avoid bugs, like types 18:07 <+bridge> [ddnet] yep 18:07 <+bridge> [ddnet] try messing with lambdas and captured variables 18:07 <+bridge> [ddnet] that's horrible in python 18:08 <+bridge> [ddnet] `return [lambda: i for i in range(10)]` 18:08 <+bridge> [ddnet] gives you a nice list of ten functions returning 9 18:08 <+bridge> [ddnet] wait what? 18:09 <+bridge> [ddnet] the "fix" for that is: `return [lambda i=i: i for i in range(10)]`, which is even worse 18:11 <+bridge> [ddnet] who decides the direction of python now that the BDFL is gone? 18:12 <+bridge> [ddnet] the community, probably? I don't know 18:14 <+bridge> [ddnet] ah look, JS has the very same problem 18:14 <+bridge> [ddnet] `var array = 0; for (var i = 0; i < 10; i++) { array.push(function() { return i; }); }` 18:14 <+bridge> [ddnet] gives you a list of ten functions returning 10 18:16 <+bridge> [ddnet] how to fix it in JS? 18:27 <+bridge> [ddnet] you want a list of 10 functions returning from 0 to 9? 18:28 <+bridge> [ddnet] yes 18:30 <+bridge> [ddnet] `let arr = []; for(let i = 0; i < 10; i++) arr.push(() => i);` 18:30 <+bridge> [ddnet] (arrow functions (lambdas) have lexical scope) 18:32 <+bridge> [ddnet] so the fix is replacing `function() { return ...; } ` with `() => ...`? 18:32 <+bridge> [ddnet] yes 18:32 <+bridge> [ddnet] or `() => { return ...; }` if you prefer 18:35 <+bridge> [ddnet] hm, I don't get why the initial python example doesn't work 18:36 <+bridge> [ddnet] because it captures the variable i which changes later on 18:37 <+bridge> [ddnet] why does the JS example not work with functions but with lambdas? 18:37 <+bridge> [ddnet] or "what is lexical scope?" 18:37 <+bridge> [ddnet] and "what scope do `function()s` have?" 18:38 <+bridge> [ddnet] now that I've read it again, I don't think the scope was the issue here 18:42 <+bridge> [ddnet] ah an anonymous function would work too, it's scope creep 18:43 <+bridge> [ddnet] it's `let i` instead of `var i` that's fixing it 18:50 <+bridge> [ddnet] @heinrich5991 with let the scope of i is constrained to what's inside the loop body, at each iteration you get a new i, and the new anonymous function you create forms a closure around that new instance of i 18:50 <+bridge> [ddnet] with var it's scope is not limited at all so it suffers from the same issue with python 18:51 <+bridge> [ddnet] ah 18:57 <+bridge> [ddnet] js is such a minefield 18:59 <+bridge> [ddnet] but with ES6 we now have better control over scope (like let)