DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2021-02-24 00:00:00Z and 2021-02-25 00:00:00Z

i dont think anyone knows sql here better than u

08:04

justatest

Hey, I need some help, I think. I have 2 issues, first one is this crash: https://pastebin.com/raw/CzcgPjRr And second one is this: https://github.com/fokkonaut/F-DDrace/blob/F-DDrace/src/game/server/entities/trail.cpp#L121-L122 if i uncomment this line it crashes due to out of bounds (historypos + 1), but i dont know a good fix for it since I cant reach timakro who originally made the trail

@Learath2 If you're considering pass, at least for android you wouldn't have to write your own app. I use android password store It's awesome

timakro: @fokkonaut was looking for you

It's too long ago that I wrote that code, I don't know any more about what I did there than you do @fokkonaut

ouch

ouch

game/collision.cpp : 1093

        dbg_msg("collision", "something is VERY wrong with the %s layer please report this at https://github.com/ddnet/ddnet, you will need to post the map as well and any steps that u think may have led to this", aBuf);

nice error message lol

got that couple of times

still from ddrace times where we had the random tile bug

12:40

that one was very annoying

12:40

it caused random tiles to appear in the map, but only for the server

12:41

so then you'll find you're 90% through a map, and the server has decided to spawn a blocker at just the right place to block you from finishing

12:41

or a random death tile

12:41

(not seen in entities btw)

good that you fixed it

idk who fixed it, probably GreYFoX

Sorry if this isnt the proper channel but who should i ask to have a flag added

14:32

It would be nice for Québec people to find each other if we had our flag

Isnt quebec a city?

14:44

I think the idea is to only support countries. So you would search for people with a canadian flag

yeah exactly lets docs ourselfs to random people on the internet

@NeXus Quebec is a province in canada, the main point is that quebec is a french province so it would help identify french speakers

[ddnet/ddnet] Issue opened: #3639 Suggestion: Add Quebec flag

From Discord: >[3:32 PM] M Emile: Sorry if this isnt the proper channel but who should i ask to have a flag added >[3:32 PM] M Emile: It would be nice for Québec people to find each other if we had our flag >[3:40 PM] NeXus: Isnt quebec a city? >[3:44 PM] NeXus: I think the idea is to only support countries. So you would search for people with a canadian flag >[4:02 PM] Ama: yeah exactly lets docs ou...

@Learath2 let's deploy now?

Can you also commit or pr a bump in the version cutoff for the new messages so we can deploy master?

add Tatarstan flag too xdd

comment on github

16:02

@Learath2 bump to which version?

99999 I guess, we can bump it down when releasing 15.4 with the new messages

hm

Or 15040 if you want to make sure we don’t forget as it’ll break

[ddnet/ddnet] Pull request opened: #3640 Bump version for the new protocol to DDNet 15.4

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...

Taking one last look over the mysql pr

16:09

Huh, why did you cancel that? Doesn’t fix the issue?

@Learath2 forgot that I had <= instead of <

16:10

fixed it now

Um, <= sounds correct to me

depends on whether VERSION_DDNET_MSG_LEGACY is on the right or the left side

Oh yeah

@heinrich5991 I spotted another assert

16:26

m_InUse

I guarantee that all my other code also relies on asserts firing

will fix

I mean we will compile this in release mode, which means it won’t be thread safe really. That’s why I’m concerned

there's only one thread (I do crash on a failed assert after all)

16:28

(I assumed a crash on a failed assert after all*)

16:33

@Learath2 fixed

16:33

oh, boost

[ddnet:master] 2 new commits

794d768 Bump version for the new protocol to DDNet 15.4 - heinrich5991 78041ba Merge #3640 - bors[bot]

16:34

[ddnet/ddnet] Pull request closed: #3640 Bump version for the new protocol to DDNet 15.4

Whatever it’s only on CI

I pushed 7 seconds before your last review, I could have sneaked in code

16:36

(removed boost)

Lol

maybe bors should decline to do its job if the latest push is <1min ago?

16:38

suggest braces around empty if block

16:38

fine..

16:39

I wonder why I forgot the braces there anyway

That CodeQL step takes some time

[ddnet:master] 4 new commits

d7019a2 Get rid of the MySQL C++ connector as a dependency - heinrich5991 d37b62e Get rid of unused IN_USE status for db connections - heinrich5991 9568d10 Remove exceptions from database code - heinrich5991 2fb1c72 Merge #3568 - bors[bot]

16:53

[ddnet/ddnet] Pull request closed: #3568 Get rid of the MySQL C++ connector as a dependency and disable C++ exceptions

so, now let's deploy it on ger2, and if not everything catches fire, deploy it everywhere?

16:55

@Learath2 I'll do that, okay?

Lets deploy it on ger2, restart a single server and finish a map

16:55

That sounds a good enough test

@Learath2 8313

@heinrich5991 I'll get back to doing some math, lmk if you need anything

ye

17:11

works locally obviously

17:11

wouldn't be fun otherwise

Could it be that many servers are connected to the same mysql server and you need some magic flag to get it to work in parallel?

I don't htink so

17:12

I had that error in testing as well

17:13

it happened when I wasn't correctly finishing processing the results

Maybe you still have a result lying around?

oh wait

You could try wrapping MYSQL_RES in a smart object aswell to make sure it gets freed even if we don't use all of it when it goes out of scope

I reproduced sth locally

does master server support 15.3 clients already?

@deen should but now master has broken mysql

18:16

not should, it does we checked

Can I not bind gfx_fullscreen?

18:25

I did: bind ctrl+f5 gfx_fullscreen 1

18:27

Ok it works

what was the cmd for deepfly ?

18:29

the cmd that make that when u shoot ur dummy shoot

For standard deepfly you can just use the dummy hammerfly command normally

18:30

If you want to time your shots the deepfly bind is bind x "bind mouse1 \"+fire; +toggle cl_dummy_hammer 1 0\""

brownbear

18:33

i am very bad bcs dont work lmao

18:34

i mean he doesnt hit me, he hit but deep

If you are trying it on the new map it has been disabled

@timakro could you at least take a look at the backtrace and guess what could be changed?

im quite the guessing type of guy

@fokkonaut I'd guess in line 120 of trail.cpp the index HistoryPos is out of bounds?

yea

19:26

but i dont know what a good fix for this is

19:26

and it doesnt happen always xd

Looks like floating point inaccuracy

19:28

I don't know the reasoning for having an extra m_TrailHistoryLength variable that needs to be kept in sync with the data in the deque

Try saving a little extra history

19:41

Maybe just to be 100% sure break before accessing an invalid indec

19:41

The line you commented was for interpolating between two history points

19:42

Might look fine without

[ddnet/ddnet] Pull request opened: #3641 Fix errors introduced by the MySQL C API pull request

For the first one, I'm not sure how this ever worked (the ranks did end up in the DB though). The other one was introduced by a rebase.

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect ex...

@Learath2 ^

I'd be curious how the code looked before your edit

@timakro https://github.com/DDNetPP/DDNetPP/blob/master/src/game/server/entities/character.cpp#L1854-L1934

DDNetPP/DDNetPP

Based on DDraceNetwork by deen & contributors (www.ddnet.tw) which is based on the great game teeworlds (www.teeworlds.com). - DDNetPP/DDNetPP

unrelated, @Learath2: you said you worked with swagger before. does one have to design one's API with that tool in mind, or can you describe the API with it later on?

[ddnet:master] 3 new commits

83da81e Use ExecuteUpdate for SQL INSERT statements - heinrich5991 44437ee Fix return values for loading saves - heinrich5991 f006eee Merge #3641 - bors[bot]

19:56

[ddnet/ddnet] Pull request closed: #3641 Fix errors introduced by the MySQL C API pull request

@heinrich5991 I only ever used it to design an api, but I think there are tools that generate it from php code

20:18

(Using annotations)

heinrich5991

unrelated, @Learath2: you said you worked with swagger before. does one have to design one's API with that tool in mind, or can you describe the API with it later on?

I used swagger and openapi even wrotey own codegen. What I got to say - it's pretty shitty permissive tool without any contracts in mind

21:00

And all codegenerators for it are shitty generating degenerative code that doesn't work / poorly designed

21:00

If you ask me what to use I'd say use gRPC

@timakro https://pastebin.com/raw/JrVBkyq4

@timakro could you look into it again maybe? i am trying to fix it but i cant reproduce the bug at all, it just happens randomly sometimes

If its so seldom just break the loop before accessing an out of bounds value

22:16

@fokkonaut ^

https://github.com/fokkonaut/F-DDrace/commit/9cea7be036e12959102ef8992f65f726c891da25

22:17

like this?

22:18

it may look weird if i break at certain places

22:19

this seems to work, but i also dont understand what you did with AdditionalLength, since I am unable to use it because HistoryPos + 1 will instantly crash so i worked around it which seems to work nicely

No I mean just check if HistoryPos >= m_TrailHistory.size() before indexing

22:19

And break in that case

isnt that what i do now?

22:21

i think i am doing exactly thats

22:22

a break would only break the while loop which would still index at the bottom

22:22

so return is okay since nothing more is happening after the loop anyways

22:22

no?

Yes, you're right. Though there's still a possible path to a out of bounds index - maybe?

22:23

Wow the code prior was completely borked

xD

So this worked?

i dont know

22:25

it does not happen alwayys

22:25

time will tell me whether this works^^

If it still happens just add a if directly before the indexing

kk, thanks

Or rewrite this mess ^^

xD

[ddnet/ddnet] Pull request opened: #3642 Forgot to actually send Sv_RecordLegacy to old clients

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...

[ddnet:master] 2 new commits

35dca7e Forgot to actually send Sv_RecordLegacy to old clients - heinrich5991 3d86c76 Merge #3642 - bors[bot]

22:58

[ddnet/ddnet] Pull request closed: #3642 Forgot to actually send Sv_RecordLegacy to old clients

Exported 146 message(s)