Marc Durdin

> _**If**_ the keystrokes are reasonably and consistently placed. I dunno. I think it's reasonable for the end user to assume that having typed `problem`, they'll be offered `problems`, even...

We'll need some more detail including Keyman version and screenshots I think!

> What is the modelling layer error? Lexical Models. You can ignore it. Being addressed by #7037.

Please include text rather than screenshot of stack trace (so we can search it in future!) ``` TypeError: Cannot read properties of null (reading 'layout') at buildAlternates(node_modules/@keymanapp/input-processor/src/text/inputProcessor.ts:270:30) at processKeyEvent(node_modules/@keymanapp/input-processor/src/text/inputProcessor.ts:171:56) at...

Good analysis. Preloading may be a simple enough mitigation -- although possibly not desirable for very complex keyboards such as Vietnamese Telex or Hieroglyphics. > I can easily provide a...

We'll aim to get this into 13.0; given it is not a bug we need to focus on addressing bugs at this time for 12.0.

Note that there is an outstanding design issue around Alt/Option and Ctrl on macOS, not limited to web platform. See #734, #875, #5721.

Can confirm this is still a problem in 15.0.265. Am testing with the attached keyboard: [no_serial_key.zip](https://github.com/keymanapp/keyman/files/9087455/no_serial_key.zip) ``` + [SHIFT LCTRL LALT K_1] > '½' '½' + [SHIFT LCTRL LALT K_1]...

> I think we should be fine without user testing... though I could see an argument for a mild bit of user testing for predictive text in the mobile apps...

> What did you have in mind for this? Do we have access to suitable devices for such tests? I _assume_ that we'd need physical devices to test this; after...