Hi there, we’ve been using Kolibri Studio for several months and have spotted a few bugs.
When copying assets to the clipboard, several copies are sometimes created. e.g. I select a Khan Academy video, click Copy to Clipboard, and now there are two or even more copies on the clipboard, requiring me to delete superfluous copies.
When opening a folder, load time has slowed down significantly as of a few weeks ago. Oftentimes it hangs completely and I have to refresh the page. This can mean that when drilling down into three or more folder levels I have to refresh the browser with each click. Incredibly time consuming, bordering on unusable.
I get sync errors a lot of times when dragging content from the clipboard into a channel.
Bugs have been replicated across multiple machines.
Regarding the extra copies on your clipboard, it would be very helpful for debugging if you can capture a video showing the bug you’re seeing, i.e. a video of an empty clipboard, then through a series of steps, be able to replicate the clipboard bug you’re seeing.
Thank you! I’ll be investigating this today and the rest of the week.
I did some more testing, and it seems that opening folders with a lot of resources in them is what causes the slowdown. I assume that each click asks the server to send metadata on all resources in child directories, rather than just the data necessary to show the next level down.
In other words, trying to open the Math folder under Khan Academy probably asks the server to send over metadata for 21,833 resources rather than just 20 for the directory level immediately below. I may be wrong, but this would explain why load times speed up the deeper down the hierarchy I go.
It’s not quite that - the issue is that each node is being annotated with details about the descendant nodes, so the further down the hieararchy you get, the less annotation of data is happening within the API call.
We are cogniscent of these issues and the fact that some parts of the Studio backend are not implemented optimally. We are currently engaged in a full scale refactor of the frontend which will also allow us to simplify our backend implementations accordingly, but this takes some time of course!
In the mean time, these kind of reports are definitely helpful for us to do interim tweaks to improve performance.
@eriklonnroth, one of the solutions we’re trying is to set up a studio frontend instance that’s closest to you as we can get. Can you go to this website and tell me the top 3 locations with the lowest latency: http://www.gcping.com/
From there we can set up a cluster that has the lowest latency, and that can hopefully alleviate some of the slowness you’re seeing.
Gotcha, thanks! It looks like a european datacenter (either Zurich or Germany) may improve your ping times.
In the meantime, we’ve added yet another fix – increasing our database resources so we can serve more simultaneous users. Let me know if that changes load times on your side.
