You've already forked matrix-js-sdk
mirror of
https://github.com/matrix-org/matrix-js-sdk.git
synced 2025-11-26 17:03:12 +03:00
Avoid paginating forever in private rooms
In TimelineWindow.paginate, keep a count of the number of API requests we have made, and bail out if it gets too high, to ensure that we don't get stuck in a loop of paginating right back to the start of the room.
This commit is contained in:
@@ -30,6 +30,13 @@ var DEBUG = false;
|
||||
*/
|
||||
var debuglog = DEBUG ? console.log.bind(console) : function() {};
|
||||
|
||||
/**
|
||||
* the number of times we ask the server for more events before giving up
|
||||
*
|
||||
* @private
|
||||
*/
|
||||
var DEFAULT_PAGINATE_LOOP_LIMIT = 5;
|
||||
|
||||
/**
|
||||
* Construct a TimelineWindow.
|
||||
*
|
||||
@@ -179,10 +186,14 @@ TimelineWindow.prototype.canPaginate = function(direction) {
|
||||
* even if there are fewer than 'size' of them, as we will just return those
|
||||
* we already know about.)
|
||||
*
|
||||
* @param {number} [requestLimit=5] limit for the number of API requests we should
|
||||
* make.
|
||||
*
|
||||
* @return {module:client.Promise} Resolves to a boolean which is true if more events
|
||||
* were successfully retrieved.
|
||||
*/
|
||||
TimelineWindow.prototype.paginate = function(direction, size, makeRequest) {
|
||||
TimelineWindow.prototype.paginate = function(direction, size, makeRequest,
|
||||
requestLimit) {
|
||||
// Either wind back the message cap (if there are enough events in the
|
||||
// timeline to do so), or fire off a pagination request.
|
||||
|
||||
@@ -190,6 +201,10 @@ TimelineWindow.prototype.paginate = function(direction, size, makeRequest) {
|
||||
makeRequest = true;
|
||||
}
|
||||
|
||||
if (requestLimit === undefined) {
|
||||
requestLimit = DEFAULT_PAGINATE_LOOP_LIMIT;
|
||||
}
|
||||
|
||||
var tl;
|
||||
if (direction == EventTimeline.BACKWARDS) {
|
||||
tl = this._start;
|
||||
@@ -224,7 +239,9 @@ TimelineWindow.prototype.paginate = function(direction, size, makeRequest) {
|
||||
return q(true);
|
||||
}
|
||||
|
||||
if (!makeRequest) {
|
||||
if (!makeRequest || requestLimit === 0) {
|
||||
// todo: should we return something different to indicate that there
|
||||
// might be more envets out there, but we haven't found them yet?
|
||||
return q(false);
|
||||
}
|
||||
|
||||
@@ -256,7 +273,12 @@ TimelineWindow.prototype.paginate = function(direction, size, makeRequest) {
|
||||
// token. In particular, we want to know if we've actually hit the
|
||||
// start of the timeline, or if we just happened to know about all of
|
||||
// the events thanks to https://matrix.org/jira/browse/SYN-645.
|
||||
return self.paginate(direction, size, true);
|
||||
//
|
||||
// On the other hand, we necessarily want to wait forever for the
|
||||
// server to make its mind up about whether there are other events,
|
||||
// because it gives a bad user experience
|
||||
// (https://github.com/vector-im/vector-web/issues/1204).
|
||||
return self.paginate(direction, size, true, requestLimit - 1);
|
||||
});
|
||||
tl.pendingPaginate = prom;
|
||||
return prom;
|
||||
|
||||
Reference in New Issue
Block a user