[Merged by Bors] - 0.10 Blog Post: Add ECS optimizations section#549
[Merged by Bors] - 0.10 Blog Post: Add ECS optimizations section#549james7132 wants to merge 12 commits intobevyengine:release-0.10.0from
Conversation
LiamGallagher737
left a comment
There was a problem hiding this comment.
Just some grammar and spelling things
Co-authored-by: JoJoJet <[email protected]> Co-authored-by: Liam Gallagher <[email protected]>
Co-authored-by: Liam Gallagher <[email protected]>
Co-authored-by: JoJoJet <[email protected]>
|
From what I can tell, the previous blog posts don't link back to specific pull requests. Not sure if we want to start doing that here. |
I think that it's valuable, especially for deeply technical changes like this. There's some portion of the audience who will be fascinated and want to hear about the exact details, and giving them an actual link is nice. I also think it does a much better job fairly assigning credit to the authors: being able to quickly jump to the actual work that caught your eye is super useful from the perspective of trying to evaluate skills (for hiring or otherwise). |
|
Good points. It might be good to decide on a standard style for referencing PRs, though? |
I think this topic is complicated because the set of PRs that should be shared with our users (because they are useful / interesting) don't always line up with the PRs that were the most amount of work / the most technical / etc. If we start using "linking to PRs" as a regular part of our credit system, we risk bloating the blog post with noise. And we also risk only crediting things that are "exciting, linkable via a single PR, and easy to digest". Those are the main reasons I've been avoiding doing this unless absolutely necessary. |
|
The main reason why I wanted to link the PRs was the proof in the pudding (i.e. the benchmark results) are most readily visible there. Moving them to the main blog post, particularly with how many of them there are, might be noisier than if we just linked them. Might be worth moving the link to the end of each blurb instead of at the start, where it interjects into one's consciousness more aggressively. |
|
Yeah I think its reasonably justified in this context / the current state of things is pretty good. |
|
bors r+ |
I split out the adaptive batching part from the rest of the optimizations work since that has a user-visible impact. Co-authored-by: James Liu <[email protected]>
|
Pull request successfully merged into release-0.10.0. Build succeeded: |
I split out the adaptive batching part from the rest of the optimizations work since that has a user-visible impact.