confluence
confluence copied to clipboard
GoogleChromeLabs
Confluence metrics API lists are slow to load and progress is not obvious
If I click a "failure to ship" graph datapoint and try to look at what APIs are claimed to be represented by that datapoint, I get an empty list.
If I click "expand all", the list stays empty. After some more random clicking, I can sometimes get data.
I think this is a performance problem, not functional. But sometimes it takes a very long time.
Some of these data points pull data from a large number of browsers. We have some ideas in the works for compressing the data model (which is, atm, a little verbose), but launching that change will take time.
If you're not seeing the little progress bar above the data table while you wait for data, that's a bug that can probably be fixed more quickly (but you'll still have to wait a while for the data :sad:)
@mdittmer, maybe add a toast that says to hold on after a few seconds of loading?
I didn't get that the subtly pulsing blue bar is supposed to be a progress bar (assuming that's what it is). It's not very obvious.
I didn't get that the subtly pulsing blue bar is supposed to be a progress bar
Oh. I hadn't even noticed its existence.
I think a much more obvious progress indicator is in order if the primary UI is still being loaded. ;)
Oh. I hadn't even noticed its existence.
I didn't notice it at all either until someone claimed in this issue that there was a progress bar. Without having seen this issue, I'd probably still be unaware of it.
Added the roadmap label to this. @mdittmer, what do you think is the most productive way to fix this, by improving the load time or making the loading more obvious? Will there always be some cases that take >1s to load?
For what it's worth, 1s would be pretty awesome. Right now I get a 5-10s lag after clicking "expand all" before anything shows up. And I'm on a reasonably fast (50Mbit) connection.
Yep. I think it should be possible from a raw data volume point of view, because there are about 6000 APIs, so once the list of API names is known, it should be about 750 bytes of data per browser uncompressed.
My thoughts are:
ASAP, add spinner in addition to progress bar. Progress bar will be the only thing visible when data is rendered but being updated (e.g., fetching an unfetched release's data).
On the roadmap: compress data model and make it feasible to have all data loaded in a format that is much closer to the view's way of presenting it.