Variable types in Arduino IDE

[Phono]

After years of use of the arduino IDE + Roger's core that kept evolving, I had a rather poor development environment that I would be unable to rebuild in case of a crash, due to the number of fixes and tweaks that were accumulated over time.
I have decided to build a new environment on a fresh machine, and I have installed Arduino 1.8.12 to which I added the STM32 board collection of the official repository (stm32duino/Arduino_core_STM32 from github), plus the STM32CubeProgrammer.
This worked rather easily.
Now I am starting to port my sketches and my special libraries to this environment, and I have a problem with the data types.
Previously I used libmaple_types.h as the base for declarations. It provided me with the definitions of uint8, int8, uint16, int16, uint32, and int32.
This library was obviously a legacy from the long defunct maple leaflabs environment. So, what can I do now to get these definitions? Does the STM32 core provide for this, even if the names are different?
I have checked the help file of the Arduino IDE, and I get a confusing landscape.
byte and unsigned byte are ok.
integer and unsigned integer are 16-bit long for AVR chips, but are 32-bit long for Arduino Due. What for STM32?
long integer and unsigned long integer are 32-bit long for AVR chips, but they say nothing about the Arduino Due.
Is there an official convention for the STM32 core?

[fredbox]

You can use the fixed width types without having to include any special headers:

Code: Select all

int8_t
uint8_t
int16_t
uint16_t
int32_t
uint32_t
int64_t
uint64_t

Something like

Code: Select all

#define int8 int8_t

or

Code: Select all

typedef int8_t int8;

should work to use your previous definitions.

[Phono]

Thank you, this problem is solved. However, I have yet another problem. In my sketch I massively used the __FLASH__ attribute to store string constants in flash and spare the RAM. This attribute is not accepted by the compiler any more. What is the new syntax for this?

[stas2z]

Code: Select all

const

will be enough

[fredbox]

Code: Select all

char const * msg[] = 
{
  "test string 001",
  "test string 002",
  "test string 003",
  "test string 004",
  "test string 005",
  "test string 006"
};

Sketch uses 16456 bytes
Global variables use 3532 bytes

Change the first line to:

Code: Select all

char const * const msg[] = 

Sketch uses 16456 bytes
Global variables use 3508 bytes

The strings are saved to flash in either case. The second method also saves the pointers to flash.

[Phono]

Thanks, it worked.

[mrburnette]

For reference and history, an adjunct to fredbox's answer: https://www.embedded.com/introduction-t ... -integers/

...the 1999 update to the International Organization for Standardization's (ISO) C standard (hereafter “C99”) did just that. The ISO has finally put the weight of its standard behind a preferred set of names for signed and unsigned fixed-size integer data types. The newly defined type names are:

8-bit: int8_t uint8_t
16-bit: int16_t uint16_t
32-bit: int32_t uint32_t
64-bit: int64_t uint64_t
According to the updated standard, this required set of typedef s (along with some others) is to be defined by compiler vendors and included in the new header file stdint.h .

If you're already using a C99-compliant compiler, this new language feature makes declaring each required fixed-width integer variable or register definition as straightforward as using one of the new type names.


For PC systems, C++11 compilers will likely support: https://en.cppreference.com/w/cpp/types/integer


Ray

[fredbox]

These integer types are also available for use in Roger's libmaple core.