On automated testing, I have pushed another branch to GitHub, based on the theadfix branch, that gives a choice of paste mechanism: the current way of intercepting the OS keyboard buffer remove and count routines, and the alternative way to actually simulate the keys on the keyboard being pressed. The latter is substantially slower*, but does have the advantage that it works with OS 0.1 which doesn't have a "remove character from buffer" vector to intercept. It is even twice as slow on OS 0.1 as on OS 1.20 as OS 0.1 seems to wait long for possible key bounce before processing the key being down.
The 'paste' debug command and '-paste' command line option use the OS mechanism. A new debug command 'pastek' and '-pastek' command line option invoke the keyboard-based version and the Edit menu now has a choice. Hopefully that will enable automated tests to be done on OS 0.1 models.
* The keyboard-based paste was how I first implemented it but, while it was fine for pasting in a few lines, a decent sized program took a long time to enter so the OS intercept was a way to speed up paste dramatically. Then SteveF (ZornsLemma) brought back some of the original keyboard paste code when implementing logical keyboard mode. The new keyboard paste piggy backs on that.
The 'paste' debug command and '-paste' command line option use the OS mechanism. A new debug command 'pastek' and '-pastek' command line option invoke the keyboard-based version and the Edit menu now has a choice. Hopefully that will enable automated tests to be done on OS 0.1 models.
* The keyboard-based paste was how I first implemented it but, while it was fine for pasting in a few lines, a decent sized program took a long time to enter so the OS intercept was a way to speed up paste dramatically. Then SteveF (ZornsLemma) brought back some of the original keyboard paste code when implementing logical keyboard mode. The new keyboard paste piggy backs on that.
Statistics: Posted by Coeus — Sat Feb 22, 2025 12:36 am