Core: Use long commit hash instead of abbreviating

[crueter]

Fixes #5230

Basically just stores it as an array of uint8_t.
Hopefully FEX never reaches SHA1 collisions so
this should (TM) never collide.

Signed-off-by: crueter crueter@eden-emu.dev

[bylaws]

Will need accompanying proton changes

[crueter]

Done, assuming you refer to the steam VERSIONS file

[crueter]

A potentially notable edge case: odd-numbered hash lengths don't work here:

static constexpr std::array<uint8_t, 20> GIT_HASH = {0x12, 0x34, 0x56, a};

Could be solved by padding with a singular 0 if the length is odd?

[crueter]

wait, could just use a list

[crueter]

Okay, technically it isn't "correct" per se, but it's all I think is necessary for an edge case like this

static constexpr std::array<uint8_t, 20> GIT_HASH = {0x12, 0x34, 0x56, 0xa};

[neobrain]

A potentially notable edge case: odd-numbered hash lengths don't work here:

What's the problem and how did you solve it?

[crueter]

The previous problem was caused by implanting a 0x, every two positions, which ended up leaving a dangling letter/number at the end

To solve it, I just used ; to just construct a list and then use TRANSFORM and JOIN to convert it to an array-like initializer

...and now I solved it by adding another case in the regex for that. yay!

[neobrain]

Just noting for reference that this should be tested on arm64 and that the unrelated changes should be confirmed functional as pointed out in my other comments.

[neobrain]

Thanks, looks all good now - could you squash the patches together appropriately?

[crueter]

done

[neobrain]

Re-tested this against my WIP code caching patches, all working fine. Thanks again for picking this up!