Reorder pins in bi_pins_with_names to be correct

[will-v-pi]

Alternative to #1791 - instead of asserting when the order is incorrect, reorder the names so they are in the correct order for the pin mask

Requires more pre-processor stuff, but doesn't break existing code with the pins in the wrong order, so pico-examples/524 would no longer be required

Has been tested using the following python code to generate all permutations of bi_decls which are then copied and compiled into a binary, and then checked with picotool info to ensure pins are all named correctly

import itertools

pins = [
    'PICO_DEFAULT_SPI_RX_PIN, "SPI RX"',
    'PICO_DEFAULT_SPI_TX_PIN, "SPI TX"',
    'PICO_DEFAULT_SPI_SCK_PIN, "SPI SCK"',
    'PICO_DEFAULT_SPI_CSN_PIN, "SPI CS"'
]

pinperms = itertools.permutations(pins, 2)
for p in pinperms:
    print(f"bi_decl(bi_2pins_with_names({p[0]}, {p[1]}));")

pinperms = itertools.permutations(pins, 3)
for p in pinperms:
    print(f"bi_decl(bi_3pins_with_names({p[0]}, {p[1]}, {p[2]}));")

pinperms = itertools.permutations(pins, 4)
for p in pinperms:
    print(f"bi_decl(bi_4pins_with_names({p[0]}, {p[1]}, {p[2]}, {p[3]}));")

[lurch]

ZOMG! 🤯 As long as you've tested this, I'll assume it's okay 😉

Although I guess your tests could be simplified to use e.g. pinperms = itertools.permutations(pins[:2], 2) instead?

[kilograham]

Yeah, it is a bit scary; i was wondering if we should do a recursive implementation, but

1. that means splitting out a separate macro for the string part
2. I don't know that it is significantly simpler

So unless anyone has any fantastic ideas i say we take this as is ;-)

[lurch]

So unless anyone has any fantastic ideas i say we take this as is ;-)

https://github.com/svpv/qsort 🤣

[kilograham]

I'm gonna believe you!