Automatically retry failed asset loads

[alice-i-cecile]

Do we not retry failed asset loads? I would assume they would fail, back off, and then retry.

We currently do not.

Originally posted by @alice-i-cecile in #17377 (comment)

[alice-i-cecile]

This should be configurable!

[cart]

As a note, this should be implemented in such a way that only retries things that aren't actual logic errors. Ex: we should not retry when a file is missing.

Unfortunately each class of error / context might also want different retry counts / delays / etc. Ex: a timed out HTTP request may have different retry requirements than a "too many file descriptors" error.

[brianreavis]

• See Asset Load Retry Plugin #11349

It's a little out of date with bevy/main, but I have it up-to-date on my fork and can update the PR if desired.

As a note, this should be implemented in such a way that only retries things that aren't actual logic errors. Ex: we should not retry when a file is missing. Unfortunately each class of error / context might also want different retry counts / delays / etc. Ex: a timed out HTTP request may have different retry requirements than a "too many file descriptors" error.

It does handle the above concerns.

// ...
            match read_error {
                AssetReaderError::NotFound(_) => {}
                AssetReaderError::Io(io_error) => match io_error.kind() {
                    ErrorKind::InvalidInput
                    | ErrorKind::InvalidData
                    | ErrorKind::NotFound
                    | ErrorKind::PermissionDenied
                    | ErrorKind::Unsupported => {}
                    _ => return self.retry_settings,
                },
                AssetReaderError::HttpError(status_code) => {
                    match status_code {
                        // Retry after server errors and rate limits
                        500..=599 | 429 => return self.retry_settings,
                        _ => {}
                    }
                }
// ...