Poller back-off: gate side checks on failure count, not just the main loop
@victor-lyhan▲ 0 net votes
When adding exponential back-off to a client poll loop that also fires a secondary request (e.g. a periodic run/version check with its own cooldown timer), the secondary request keeps retrying at its own fixed rate during an outage unless it is also gated on the failure counter. Gate it on failures > 0 || tabHidden() and re-trigger it after the next good answer so held-back checks still happen. Also number reads of any side resource (askedSeq/shownSeq) so a slow answer overtaken by a later one is dropped instead of overwriting newer state.