From the change:
// xtensa has more registers than an instruction can address. The 16 that
// can be addressed are called the "window". When a function is called or
// returns the window rotates. This allows for more efficient function calls
// because ram doesn't need to be used. It's only used if the window wraps
// around onto itself. At that point values are "spilled" to empty spots in
// the stack that were set aside. When the window rotates back around (on
// function return), the values are restored into the register from ram.
// So, in order to read the values in the stack scan we must make sure all
// of the register values we care about have been spilled to RAM. Luckily,
// there is a HAL call to do it. There is a bit of a race condition here
// because the register value could change after it's been restored but that
// is unlikely to happen with a heap pointer while we do a GC.
Fixes#2907
* Fix flash writes that don't end on a sector boundary. Fixes#2944
* Fix enum incompatibility with IDF.
* Fix printf output so it goes out debug UART.
* Increase stack size to 8k.
* Fix sleep of less than a tick so it doesn't crash.
Length was stored as a 16-bit number always. Most translations have
a max length far less. For example, US English translation lengths
always fit in just 8 bits. probably all languages fit in 9 bits.
This also has the side effect of reducing the alignment of
compressed_string_t from 2 bytes to 1.
testing performed: ran in german and english on pyruler, printed messages
looked right.
Firmware size, en_US
Before: 3044 bytes free in flash
After: 3408 bytes free in flash
Firmware size, de_DE (with #2967 merged to restore translations)
Before: 1236 bytes free in flash
After: 1600 bytes free in flash
Sommersoft noticed that without submodules (as RTD currently does),
the "make -qp" failed due to trying to find files within submodules.
The "print-VAR" target doesn't have any dependencies, so it sidesteps
the problem.
If we move ulab's pyi files into the submodule, though, we'll need to
fetch submodules anyway during doc building.
Revisiting this because I noticed the support matrix said that pyruler
had several modules it obviously didn't, such as the recently added
vectorio.
It is less error-prone because we can use the final values instead
of re-coding the Makefile logic in python. The only things we need
to do are invoke make in "print-database" mode, then chase any
indirect references like CIRCUITPY_VECTORIO = $(CIRCUITPY_DISPLAYIO)
It does take longer, about 45s on my laptop.
