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
How much are we concerned about server load/compute with warming/pre-streaming?
What should the default behavior be? Should this be context dependent at all? Frame size? User on wifi/mobile?
How much to warm? A ratio of total frames? Whole trajectories?
What degree of control should the front end implementation have over this behavior?
Do we want to make changes to the back end? streaming behaviors defined as "resume" or "pause" seem to presume that we stream only during playback, warming can probably be built without functional changes, but the naming may be less accurate/precise afterwards.
Control of playback speed is a requested front end feature, but currently playback speed is very much determined by network latency, at least on first play through, total/partial warming makes this a more reasonable thing to build? As does preserving cached frames when possible (multi part cache is out of scope here, but would facilitate this as well).
Sort out warming when cache size is limited.
Joe's rough draft idea 10/23:
leave existing controller methods in place, no breaking changes
don't change octopus/websocket message types (only update these after the fact if more naming accuracy is desired), use intervals and other checks internal to the viewer to manage warming
expose props/controller methods to allow front end to initiate warming
provided a default ratio of frames for warming, allow front end to update that amount down, or up to 100%
allow warming to be initiated pre-playback/time jumps, and then allow playback to proceed with old behavior, or with ongoing buffering in advance of playback position
Acceptance Criteria
viewer can request and cache some some of frames without initiating playback
if warming initiated, default ratio of frames to pre-stream is 25%
decided upon default behavior happens
decided upon implementation options/controls are exposed and function as expected
Details
Please provide any helpful specifications
The text was updated successfully, but these errors were encountered:
Use Case
To satisfy this UX ticket for a buffering indicator on the website time slider, and potentially to improve performance, and potentially to facilitate playback speed feature, cache warming (pre-streaming of frames prior to playback) we're going to warm the cache!
Open questions:
Joe's rough draft idea 10/23:
Acceptance Criteria
Details
Please provide any helpful specifications
The text was updated successfully, but these errors were encountered: