-
Notifications
You must be signed in to change notification settings - Fork 42
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Bevy 0.15 and latest ggrs main #114
Conversation
This is what bevy 0.15 migration docs suggest
Default, Serialize and Deserialize instead of bytemuck traits.
Hmmm... This leads to a panic for me:
As soon as one of the clients gets input and tries to roll back
Might be related to latest changes in ggrs? Ideas @caspark ? |
@johanhelsing this looks like the crash when rolling back while at the max prediction that gschup/ggrs#79 introduced, which then lived in master for a couple weeks until I fixed it in gschup/ggrs#88 (see 1st bullet point). Based on timing of your comment and the merging of that 2nd PR I am guessing you didn't have that fix PR's commits included when testing? If you do still get the issue with the fix PR's commits then that probably means I stuffed something up (or failed to completely fix #79 correctly), in which case I'd love to know more about it - though ideally in a non-bevy_ggrs example as I don't use bevy so haven't looked at the bevy_ggrs code at all yet :) |
Pretty sure i did cargo update today. Can't check now though |
You make it sound like I'm the evil guy here ;) |
Haha, yes @gschup , you're the bad guy for porting GGPO to Rust while figuring out a much saner design in the process, and then maintaining & improving upon it for the last 3 years, entirely for free! Only a true monster would do that. Kidding aside: just to be clear, I didn't mean to paint you as "the bad guy", so if I came across that way, sorry: I apologize for my choice of wording. I was just trying to say that I thought that particular crash wasn't due to my changes, and provide sufficient context as to why (though it is entirely possible that I have broken ggrs in other ways). |
I did a bisect, and this was caused by gschup/ggrs@b13e7fc... And then fixed again in gschup/ggrs@dc25e8d So your guesses were spot on @caspark 👍 So this means, don't hold up on the ggrs release because of this! |
I just realized that we already depend on ggrs main in bevy_ggrs main, so there is no reason not to merge this I think? We can just do another minor PR when doing the crates relases? |
Not sure if it's something to do with this thing, but I'm having this issue where I get ghosts of gltf meshes that don't appear in the world when I go over it with bevy-egui-inspector. According to someone on the bevy discord, this could be a desync between the main and the render worlds. It happens when call despawn_recursive on an entity that has a child that has a SceneRoot that points to a gltf with a few more levels of hierarchy. |
I can't speak for the changes in |
Did that, ghosts didn't show up (yay?). I did just now notice that I'm getting some logs in the console, but only when check distance is not zero:
Probably related to that. Is there like a best practice for nested entities with ggrs that I should be aware of? I mean stuff like, should I call |
Yes, the children also need add_rollback, we don't do despawn recursive when rolling back entities, so all entities need in the hierarchy to have the rollback component added bevy_ggrs/src/snapshot/entity.rs Line 71 in cdd5740
Are you not getting these issues on bevy 0.14 |
I feel like this may be on my end, moved the conversation over to discord. Would appreciate any help in my descend to madness over yonder. |
This is #113, but with the extra changes needed after gschup/ggrs#82 by @caspark got merged
[ ] Change to ggrs crates.io release once it's out