Manage confined entries in synchronization monitor

[vxgmichel]

As mentioned in this conversation, the WorkspaceOpsOutboundSyncEntryConfined event has not been implemented.

In parsec v2, it was emitted when a confined entry tries to get synchronized:

parsec-cloud/parsec/core/fs/workspacefs/sync_transactions.py

Lines 362 to 369 in d397c03

	confinement_point = await self._get_confinement_point(entry_id)
	if confinement_point is not None:
	# Send entry confined event for the sync monitor
	self._send_event(
	CoreEvent.FS_ENTRY_CONFINED, entry_id=entry_id, cause_id=confinement_point
	)
	# Do not perform the synchronization
	return None

It was linked to a handler in the sync monitor:

parsec-cloud/parsec/core/sync_monitor.py

Lines 445 to 452 in d397c03

	def _on_entry_confined(
	event: CoreEvent, entry_id: EntryID, cause_id: EntryID, workspace_id: EntryID
	) -> None:
	ctx = ctxs.get(workspace_id)
	if ctx is not None:
	ctx.set_confined_entry(entry_id, cause_id)
	# No need to call `_trigger_early_wakeup` here given due time is not updated
	# (by definition, a confined entry doesn't have to be synced)

Which allowed the sync monitor to keep track of the confined entries and their respective cause.

This information was then used here:

parsec-cloud/parsec/core/sync_monitor.py

Lines 176 to 188 in d397c03

	def set_local_change(self, entry_id: EntryID) -> bool:
	# Ignore local changes in read only mode
	wake_up = False
	if self.read_only:
	return wake_up
	# Pop confined entries related to current entry_id
	local_confinement_points = self._local_confinement_points.pop(entry_id, ())
	# Tag all confined entries as potentially changed
	for confined_entry in local_confinement_points:
	if self.set_local_change(confined_entry):
	wake_up = True

This way, all the confined entries corresponding to a given cause would get re-integrated into the synchronization system when the cause gets updated.

[mmmarcos]

According to @vxgmichel this still needs to be implemented, but it is not a priority since #8068 was fixed by #8192.