beachball when creating a new window
- iTerm2 version: 20180803-nightly
- OS version: 10.14 beta (18A347e)
2017 15" MBP with a Radeon 555 and Intel graphics on AC power so I'm assuming the discrete GPU is being used
For every window I open after the first one, a new window opens but no text shows up in it and if I hover over that window, the beachball shows up until I click on another window inside the app. Then the beachball clears, and the shell prompt finally shows up in the newly-opened window.
Detailed steps to reproduce the problem:
- Start app
- Cmd-n to make new window (just the "default" login shell one, I haven't tried with any other kind of window)
- Window pops up but there's no text
- Click in another already open window inside the app
- See new window now has text in it and is responding (no more beachball)
It may be possible that just moving the mouse cursor outside of any app windows causes it to un-wedge.
When I look at the debug log there's a gap in the timestamps (between line 5551 and 5552 in the attached one):
1533326771.941624 iTermMetalDriver.m:1706 (-[iTermMetalDriver complete:]): finished draw with group of frame <iTermMetalFrameData: 0
x7fe79a0ff400 age=0.011285 frameNumber=278/280 status=retired>
1533326773.834638 iTermMetalDriver.m:1681 (-[iTermMetalDriver complete:]): Completed
1533326773.834672 iTermTextRendererTransientState.mm:827 (-[iTermTextRendererTransientState didComplete]): BEGIN didComplete for <iTermTextRendererTransientState: 0x7fe79aa45800>
1533326773.834679 iTermTexturePageCollection.h:93 (prune_if_needed): Not pruning
1533326773.834683 iTermTextRendererTransientState.mm:829 (-[iTermTextRendererTransientState didComplete]): END didComplete
1533326773.834700 iTermMetalDriver.m:1689 (-[iTermMetalDriver complete:]): Recording final stats
1533326773.836452 iTermPreciseTimer.m:333 (iTermPreciseTimerLog): -- Precise Timers for [driver 5]: Metal Frame Data --
Statistic µ N p50 p75 p95 [min distribution max]
------------------------------------ ------- ----- ------- ------- ------- ------------------------
😱 endToEnd 526.2ms 4 100.2ms 1900.8ms 1900.8ms [10.54ms █ 1900.79ms]
| 😱 gpu 477.2ms 4 7.2ms 1893.8ms 1893.8ms [3.69ms █ 1893.83ms]
| | 😱 scheduleWait 475.9ms 4 7.2ms 1890.5ms 1890.5ms [1.52ms █ 1890.55ms]
| | 🙁 dispatchToPQ2 0.8ms 4 0.8ms 2.2ms 2.2ms [0.03ms █ 2.16ms]
The activity monitor sample isn't from the same run as the debuglog run. Sorry.
This problem was happening on the non-nightly 3.2.0b9 release so I switched to the nightly. I don't have the problem on my 10.13 computer but