gfx-rs itself doesn't use any allocations on a call-by-call basis, and it would be great if gfx_text didn't impose this overhead either. The allocations come from the constructed meshes and batches, even if implicit. What can be done:
- have an
OwnedBatch created once on init
- update its
Slice with updating the vertex/index buffers that the batch uses
- send it to a stream as usual
gfx-rsitself doesn't use any allocations on a call-by-call basis, and it would be great ifgfx_textdidn't impose this overhead either. The allocations come from the constructed meshes and batches, even if implicit. What can be done:OwnedBatchcreated once on initSlicewith updating the vertex/index buffers that the batch uses