You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
According to crash stats, a new Discord SDK crash is top #2 crash in MTA on the latest nightlies.
These nightlies are really not widely distributed yet, so this is very worrying.
To reproduce
Unknown (per issue description about crash stats).
Besides that, unfortunately dumps for this crash are never complete (e.g 'Extra info' that we collect isn't included, such as memory usage when it crashed). It is possible that there was no memory available to capture that on time. Read "Additional context" down below for more notes on memory/allocations.
It seems to have been present since the very start of adding Discord integration. See below:
The low count contrast with some entries is due to lower build distribution on those nightlies.
Additional context
If we speculate on these stats (global crash stats, of any module, on r20506 which has a 1.1% share in MTA userbase):
Then we can expect the crash count running into tens of thousands, once the builds get distributed beyond nightlies.
We have enough 0xE06D7363 (usually memory or allocation related) crashes on MTA already, so that would be really bad. If memory related at all, it's probably a result of a mem leak in the Discord integration itself, since it's new and doesn't come in the place of another existing one.
I already put question marks on overall Discord integration stability in the recent past, and likened it to how unstable CEF was in the beginning, making MTA unstable. I also expect other crashes on core.dll to be Discord SDK related, like was in my original remark. But not to discuss that here, we might need a separate issue to keep track of these problems in 1 place.
I have wrote my discord integration in such a way that it is totally easy to switch back to discord-rpc so just say the words, we'll get rid of trash from MTA repo in an instant.
Due to issue #1370 this crash might disappear once we migrated (if it was in the Discord SDK or related to how MTA integrated Discord SDK), but not if the problem is located in a part of the integration that will persist (which is very likely due to the above quote), it's important to keep track of this problem and make sure it's resolved even on discord-rpc.
I also have doubts on whether or not having this while still on Discord SDK is acceptable; the ETA for discord-rpc migration is MTA 1.6 (see #1370 (comment) and having an unstable nightly for a few months sounds like a bad idea. Maybe reverting Discord SDK entirely and waiting for discord-rpc to complete is a better idea, we will be freed of licensing conflicts and can have a fresh look on the stability issues once migrated.
Note: I am only speculating it's memory or allocation related, but it crashes in discord_game_sdk!rust_eh_personality+0x2295a and I am unable to investigate it, so disregard these remarks if you find otherwise.
The text was updated successfully, but these errors were encountered:
Describe the bug
According to crash stats, a new Discord SDK crash is top #2 crash in MTA on the latest nightlies.
These nightlies are really not widely distributed yet, so this is very worrying.
Dumptrace:
To reproduce
Unknown (per issue description about crash stats).
Besides that, unfortunately dumps for this crash are never complete (e.g 'Extra info' that we collect isn't included, such as memory usage when it crashed). It is possible that there was no memory available to capture that on time. Read "Additional context" down below for more notes on memory/allocations.
Version
Build r20462 (850c76d) and up
It seems to have been present since the very start of adding Discord integration. See below:
The low count contrast with some entries is due to lower build distribution on those nightlies.
Additional context
If we speculate on these stats (global crash stats, of any module, on r20506 which has a 1.1% share in MTA userbase):
Then we can expect the crash count running into tens of thousands, once the builds get distributed beyond nightlies.
We have enough 0xE06D7363 (usually memory or allocation related) crashes on MTA already, so that would be really bad. If memory related at all, it's probably a result of a mem leak in the Discord integration itself, since it's new and doesn't come in the place of another existing one.
I already put question marks on overall Discord integration stability in the recent past, and likened it to how unstable CEF was in the beginning, making MTA unstable. I also expect other crashes on core.dll to be Discord SDK related, like was in my original remark. But not to discuss that here, we might need a separate issue to keep track of these problems in 1 place.
quoting @0x416c69:
Due to issue #1370 this crash might disappear once we migrated (if it was in the Discord SDK or related to how MTA integrated Discord SDK), but not if the problem is located in a part of the integration that will persist (which is very likely due to the above quote), it's important to keep track of this problem and make sure it's resolved even on discord-rpc.
I also have doubts on whether or not having this while still on Discord SDK is acceptable; the ETA for discord-rpc migration is MTA 1.6 (see #1370 (comment) and having an unstable nightly for a few months sounds like a bad idea. Maybe reverting Discord SDK entirely and waiting for discord-rpc to complete is a better idea, we will be freed of licensing conflicts and can have a fresh look on the stability issues once migrated.
Note: I am only speculating it's memory or allocation related, but it crashes in
discord_game_sdk!rust_eh_personality+0x2295a
and I am unable to investigate it, so disregard these remarks if you find otherwise.The text was updated successfully, but these errors were encountered: