Limits on Babylon.js AssetsManager retry attempts and duplicate write prevention

Question

The goal is to confirm whether Babylon.js AssetsManager’s built‑in retry mechanism guarantees that an asset’s onSuccess callback is invoked exactly once when developers supplement the default retries with custom calls to task.retry().

Constraints arise because the internal retry counter is not exposed in versions before 5.0, and the interaction between the manager’s maxRetries limit and manual task.retry() calls is not fully documented, leaving uncertainty about whether the state reset could cause the success handler to fire more than once.

Does calling task.retry() after the internal retry counter reaches the maxRetries limit still prevent duplicate onSuccess invocations? If maxRetries is changed during loading, how does the manager reconcile the existing counter? Are there conditions under which onSuccess could execute more than once despite the internal state checks?

Tasadduq Burney · Answer

Answer
When you call task.retry() after the internal retry counter has already reached the task’s maxRetries value, the AssetsManager’s guard prevents any further retry attempts and guarantees that the onSuccess callback will fire only once for that asset. The guard works by checking the counter before scheduling a new load; if the counter is equal to or greater than maxRetries, the call is ignored. Therefore, duplicate onSuccess invocations will not occur under normal circumstances.

How the counter behaves with dynamic changes

Changing AssetsManager.maxRetries mid‑load: The counter that the task already has is not retroactively updated. New task.retry() calls will see the new limit, but the counter that was already incremented remains unchanged.
  Resetting the counter manually: If a developer explicitly sets task.retries = 0 (or similar), the guard will treat the task as fresh, and subsequent task.retry() calls can succeed again. This is the only documented scenario where onSuccess could fire more than once for the same load cycle.
  Re‑requesting the same asset after a successful load: If the asset is removed from the manager’s cache and a new load task is created, the callback will fire again, but this is a new task, not a duplicate of the original.

Practical steps to verify the guard is working

Set AssetsManager.maxRetries to a known value (e.g., 2).
  Start a load task that will fail (use a wrong URL).
  After the task reaches the retry limit, call task.retry() again.
  Observe that no new network request is made and the onSuccess callback is not invoked.
  Optionally, log task.retries (if the property is exposed) to confirm it equals the limit.

When duplicates can still happen

Manual reset of the retry counter.
  Creating a brand‑new task for the same URL after a previous successful load.
  Using a custom loader that bypasses the internal guard.

Missing diagnostic detail
To give a version‑specific recommendation, could you let us know which Babylon.js release you’re targeting (e.g., 5.3, 6.0, or a pre‑5.0 build)? The internal guard was tightened in 5.0, and older releases may allow task.retry() to succeed even after the counter hits maxRetries if the counter isn’t exposed.

Limits on Babylon.js AssetsManager retry attempts and duplicate write prevention

1 answer

Answer

How the counter behaves with dynamic changes

Practical steps to verify the guard is working

When duplicates can still happen

Missing diagnostic detail

Your perspective belongs here.

0 question comments