On Windows, file system events are delivered multiple times. I expect this is due to the underlying technology.
However, wouldn't it be opportune to de-duplicate or at least group together these events? If the events are grouped in the same burst, they can be effectively manually de-duplicated. That would also be possible (albeit more arduous) if a timestamp for the change was exposed.
nsfw enforces a configurable "debounce" interval (default: 500ms) which basically means that bursts of fs events are delivered once per timer run.
watcher has a pollingInterval that seems similar, but fails to make any difference in my experience.
I think the difference might be that when the polling interval is over, watcher eagerly delivers events, whereas nsfw waits for an interval duration after received a new filesystem event (after idling).
Would it be conceivable to add such a debounce interval, or at least to expose the event's timestamp? Adding an automatic de-duplication parameter would be even better.
On Windows, file system events are delivered multiple times. I expect this is due to the underlying technology.
However, wouldn't it be opportune to de-duplicate or at least group together these events? If the events are grouped in the same burst, they can be effectively manually de-duplicated. That would also be possible (albeit more arduous) if a timestamp for the change was exposed.
nsfw enforces a configurable "debounce" interval (default: 500ms) which basically means that bursts of fs events are delivered once per timer run.
watcher has a
pollingInterval
that seems similar, but fails to make any difference in my experience.I think the difference might be that when the polling interval is over, watcher eagerly delivers events, whereas nsfw waits for an interval duration after received a new filesystem event (after idling).
Would it be conceivable to add such a debounce interval, or at least to expose the event's timestamp? Adding an automatic de-duplication parameter would be even better.