Qingtian
Qingtian
@hiiamboris The hard part of modifying the hash is that we need to update the `index` of the moved items. It used some complicated code to update them (only the...
> But can you at least fix change? I can optimize cases where no items index changed.
Yes. The current implementation is quite naive and... slow. I want to see if it really causes an issue in real projects. Most of them are small projects as the...
@dockimbel Yes. `remove`, I overlooked it. It still uses the old buggy code. I'll fix it.
`unview` will always exit an event loop, while `view/no-wait` didn't create one. That's why some windows are not closed. The view engine cannot help in this case, as the user...
> Why would do-events terminate while some windows are still open? `do-events` is not smart enough. If we `view` two windows, two event loops (two do-events) launched. Then we close...
``` >> round/to 234.567% 1% == 2.3457 ;) expected 234% ``` This logic was introduced by this fix: https://github.com/red/red/commit/45043569f3fa9337fb9c4067eb3461f1c95b0377
There are two issues: 1. Should the result type coerced to `scale`? ``` >> round/to 86400 1:0:0 == 24:00:00 >> round/to 234.567 100% == 235.0 ;-- should it be 23500%...
The bug in hash! in this issue has been fixed. The `/skip` issue is design related. See #5119.
Tested on Armbian_22.08.0_Aml_s905l3a_bullseye_5.15.62, it works fine, no crashes.