next
$change_stream = $collection->watch(...);
$change = $change_stream->next;
Waits for the next change in the collection and returns it.
Note: This method will wait for the amount of milliseconds passed as "maxAwaitTimeMS" to "watch" in
MongoDB::Collection or the server's default wait-time. It will not wait indefinitely.
get_resume_token
Users can inspect the "_id" on each "ChangeDocument" to use as a resume token. But since MongoDB 4.2,
"aggregate" and "getMore" responses also include a "postBatchResumeToken". Drivers use one or the other
when automatically resuming.
This method retrieves the same resume token that would be used to automatically resume. Users intending
to store the resume token should use this method to get the most up to date resume token.
For instance:
if ($local_change) {
process_change($local_change);
}
eval {
my $change_stream = $coll->watch([], { resumeAfter => $local_resume_token });
while ( my $change = $change_stream->next) {
$local_resume_token = $change_stream->get_resume_token;
$local_change = $change;
process_change($local_change);
}
};
if (my $err = $@) {
$log->error($err);
}
In this case the current change is always persisted locally, including the resume token, such that on
restart the application can still process the change while ensuring that the change stream continues from
the right logical time in the oplog. It is the application's responsibility to ensure that
"process_change" is idempotent, this design merely makes a reasonable effort to process each change at
least once.