GitLab

It's more robust to have the version defined statically in a header file,
rather than dynamically generating it via git using a git tag.  In case
git doesn't exist, or a different source control tool is used, it's
important to still have the uPy version number available.

The new option MICROPY_HW_SDCARD_MOUNT_AT_BOOT can now be defined to 0 in
mpconfigboard.h to allow SD hardware to be enabled but not auto-mounted at
boot.  This feature is enabled by default to retain previous behaviour.

Previously, if an SD card is enabled in hardware it is also used to boot
from.  While this can be disabled with a SKIPSD file on internal flash,
this wont be available at first boot or if the internal flash gets
corrupted.

The older "bool has_finaliser" gets recast as GC_ALLOC_FLAG_HAS_FINALISER=1
so this is a backwards compatible change to the signature.  Since bool gets
implicitly converted to 1 this patch doesn't include conversion of all
calls.

Both mp_type_array and mp_type_memoryview use the same object structure,
mp_obj_array_t, but for the case of memoryview, some fields, e.g. "free",
have different meaning.  As the "free" field is also a bitfield, assume
that (anonymous) union can't be used here (for the concerns of possible
compatibility issues with wide array of toolchains), and just add a field
alias using a #define.  As it's a define, it should be a selective
identifier, so use verbose "memview_offset" to avoid any clashes.

If you happen to only have a really simple frozen file that doesn't contain
any new qstrs then the generated frozen_mpy.c file contains an empty
enumeration which causes a C compile time error.

GNU readline support for the unix port was removed in
acaa30b6 and in
5e83a75c, so it's also no longer supported
in the windows port.

To support deplop-openocd on target boards that use TEXT0_ADDR only and
have their firmware in a single binary image.

Following an equivalent fix to py/bc.c.  The reason the incorrect values
for the opcode constants were not previously causing a bug is because they
were never being used: these opcodes always have qstr arguments so the part
of the code that was comparing them would never be reached.

Thanks to @malinah for finding the problem and providing the initial patch.

All 4 opcodes that can have caching bytes also have qstrs, so the test for
them must go in the qstr part of the code.  The reason this incorrect
calculation of the opcode size did not lead to a bug is because the caching
byte is at the end of the opcode (byte, qstr, qstr, cache) and is always
0x00 when saving/loading, so was just treated as a single byte no-op
opcode.  Hence these opcodes were being saved/loaded/decoded correctly.

Thanks to @malinah for finding the problem and providing the initial patch.

On by default, turned off for minimal/bare-arm. Saves 144 bytes on x86.

Due to new webpages at nordicsemi.com, the download links
for Bluetooth LE stacks were broken.

This patch updates the links to new locations for the current
targets.

The docs are now built as one for all ports.

A static UART is useful for internal peripherals that require a UART and
need to persist outside the soft-reset loop.

To provide a cleaner and more abstract C-level interface to the UART.

The HAL just clears UE and then clears all the UART control registers.

This UART_HandleTypeDef is quite large (around 70 bytes in RAM needed for
each UART object) and is not needed: instead the state of the peripheral
held in its registers provides all the required information.

mp_obj_new_exception_msg() assumes that the message passed to it is in ROM
and so can use its data directly to create the string object for the
argument of the exception, saving RAM.  At the same time, this approach
also makes sure that there is no attempt to format the message with printf,
which could lead to faults if the message contained % characters.

Fixes issue #3004.

SHORT, INT, LONG, LONGLONG, and unsigned (U*) variants are being defined.
This is done at compile using GCC-style predefined macros like
__SIZEOF_INT__.  If the compiler doesn't have such defines, no such types
will be defined.

Allows to get address a pointer contains, as an integer.

Based on the discussion, this special method is available unconditionally,
as converting to int is a common operation.

Instead of assuming that the method is a bytecode object, and only
supporting load of __name__, make the operation generic by delegating the
load to the method object itself.  Saves a bit of code size and fixes the
case of attempting to load __name__ on a native method, see issue #4028.