Opening visualization with space does not always work. #5223
Comments
|
Michael Mauderer reports a new 🔴 DELAY for today (2023-02-08): Summary: There is 2 days delay in implementation of the (#5223) task. Delay Cause: Bug is caused by some interleaving events in the GUI controller, which are tricky to debug and fix. Possible solutions: Write easier bugs. |
|
Michael Mauderer reports a new STANDUP for yesterday (2023-02-07): Progress: Further investigated the cause of the visualization disappearing when opening it via keyboard. Found some suspicious events that seem to be related, and I am tracking those down right now. It seems It should be finished by 2023-02-09. Next Day: Next day I will be working on the #5223 task. Continue investigating and find root cause of the bug. |
|
Michael Mauderer reports a new STANDUP for today (2023-02-08): Progress: Found that there are racy updates to the graph. We attach the visualization and await some response from the backend. While waiting, we invalidate the graph in the GUI controller due to a metadata change. As we update the internal state, we notice an inconsistent visualization (the one that is waiting to be attached) and decide to detach it. Proper fix involves. Propagating the information about the attachment status to the right place, so we can take it into account when invalidating the graph. It should be finished by 2023-02-09. Next Day: Next day I will be working on the #5223 task. Design and implement the fix. |
|
This is much worse with newest (today) nightly - it happens almost every time @MichaelMauderer |
|
Michael Mauderer reports a new 🔴 DELAY for yesterday (2023-02-09): Summary: There is 1 days delay in implementation of the Opening visualization with space does not always work. (#5223) task. Delay Cause: Bug is not caused by the issue I thought it was caused. More investigation needed. Possible solutions: Write easier bugs. |
|
Michael Mauderer reports a new 🔴 DELAY for today (2023-02-10): Summary: There is 1 days delay in implementation of the Opening visualization with space does not always work. (#5223) task. Delay Cause: Bug is not caused by the issue I thought it was caused. More investigation needed. Possible solutions: Write easier bugs. |
|
Michael Mauderer reports a new STANDUP for yesterday (2023-02-09): Progress: Designed a fix that would allow the correct information about the status of the visualization state (awaiting attachment) and the node state (no visualization yet) to be considered during the synchronization of the graph state. Implemented and tested the fix to observe that this was not fixing the behavior. Continued investigation of what is going on. It should be finished by 2023-02-10. Next Day: Next day I will be working on the #5223 task. Investigate why the fix is not working. Create a better fix. |
|
Michael Mauderer reports a new STANDUP for today (2023-02-10): Progress: Upon further investigation found that the visualization state was modified just before the synchronization by another chain of FRP events: handling of showing the FRP preview. This caused the visualizations to be closed, as the handling was time sensitive, and when there was a delay due to dropped frames when opening the visualisation, then the events would cause the visualisation to be closed as a preview. Implemented a safeguard against this that looks at the number of frames that have passed in addition to the time that has passed. It should be finished by 2023-02-10. Next Day: Next day I will be working on the #5223 task. Look at updating the improved component browser entry sorting. |
|
QA: @vitvakatu |
Visualizations closing right after opening was caused by the GUI being unresponsive during loading of some visualizations. This caused the timer for measuring the time between space bar press and space bar release to be inflated. The delayed events triggered the "visualization preview mode”, thus closing the visualization has it seemed that the space bar was held down, even though the events just arrived with some delay. The problem is mitigated by considering the number of frames that have passed between the space and down and the space bar up event, instead of just the wall clock time. If the number of frames is too low, this indicates that frames were dropped to the time is inflated. Fixes #5223
|
I noticed that first |
|
Do we know why they time out? Timeouts do sound like an issue on the engine side, though. Unless the IDE is making weird requests. In either case, we might want to file an issue and investigate. |
|
@MichaelMauderer Good point. If I can reproduce this consistently I will investigate (seems like it is happening less often). |
This task is automatically imported from the old Task Issue Board and it was originally created by Adam Obuchowicz.
Original issue is here.
https://discord.com/channels/401396655599124480/1064967872012615872
There is no reliable reproduction scenario, but here are some clues:
ide watch --skip-wasm-opt.The text was updated successfully, but these errors were encountered: