A check for interruption in ConsoleController::printText caused script
to immediately stop when launched after an interrupted script.
This check was used to break out of infinite print loop, but now
becomes redundant with how micropython_port_vm_hook_loop was changed.
Change-Id: Ifa8d415e1b2c2406ad67300eb14ce46889af296f
Keyboard interruption used to be checked once every 20000 calls to
micropython_port_vm_hook_loop. However, if costly functions were
executed in between calls to this method, the delay for activating
interruptions would increase.
Now, keyboard interruption is checked after a fixed amount of time has
passed. This way, if the process waits a long time between two calls
to micropython_port_vm_hook_loop, it is still interrupted in a timely
manner.
Change-Id: I37ca3bd4a996fa086078f504340dd857526e356a
Additional checks for interruptions were making the kandinsky module
slower.
/!\ Some scripts are now very difficult to interrupt
Change-Id: I4c18273d8895deaac68084411a52556c8459d52b
When computing the barycenter between two colors, the some of the two
factors a and (1-a) was equal to 255/256 instead of 1.
Change-Id: Ia9a779d43470ef42d9430ad730e842da0f007140
KDColor::blend used to produce different colors when blending two
identical colors (ex : use draw_string in Python to print white text on
a white background).
blend now escapes early when its two color arguments are identical.
Change-Id: I01dc5a0d5e4e6a20e09fee0f346dafc313dae97b
If an expression hasUnits and is then reduced, it might not have units
anymore for instance if it was replaced with undefined).
Scenario: Enter "[5000000000000000]^20 _s" in the calculation app
Without this change, we would get weird "reduced" expression, such as
multiplication(undef, _s), but we do not always check
sSimplificationHasBeenInterrupted afterwards.
computed when the calculation is added to the store and don't change afterwards.
Otherwise, if their heights change when scrolling (due to a modification of the
display output type - ExactOnly, ApproximateOnly...), it generates crashes.
calculation height memoization. They're computed from the layouts which
don't depend on the complex format (or any other settings parameters
which could have changed).
deleting row
This fixes the following bug: input 1, OK, 1.2 OK, delete the row whose
input is '1' with the selection on the output view. The output of 1.2
does not toggle and it should.
