Fixed clamping order count in trading inventory

[l29ah]

Summary

Bugfixes "Fixed clamping order count in trading inventory"

Purpose of change

fixes #75742

Describe the solution

Argument order and types were all wrong. Now they're right.

Describe alternatives you've considered

Keeping the yet-silent UB.

Testing

Tested on the saved game provided in the issue report.

Additional context

Found due to -D_GLIBCXX_ASSERTIONS, consider enabling it for experimental builds maybe?

[l29ah]

Ouch, i didn't expect we have non-POSIX targets. Could you advise me on the type to use instead of ssize_t?

[andrei8l]

ssize_t doesn't make sense in that context and you haven't justified that change either. Even size_t is questionable, but please stick to just fixing the argument order for this PR.

[l29ah]

Fixing argument order alone won't make it work good since an unsigned integer overflow also happens here.

[andrei8l]

Fixing argument order alone won't make it work good since an unsigned integer overflow also happens here.

Irrelevant because size_t is an unsigned type and both overflow and underflow for unsigned types are well defined and the clamp would still correctly produce a number between 0 and max even if you somehow managed to select 18 quintillion items. Even if that weren't the case, ssize_t is the wrong type - you can think of it as an std::optional<size_t> rather than a sort of size_t than can hold arbitrary negative values.

[l29ah]

Clamping UINT_MAX produces another value than clamping -1, and it results in being unable to deselect items by holding - continuously. intmax_t is another option i guess.

[andrei8l]

You can also just keep it simple and revert d84cb31. It was snuck in a completely unrelated PR and the previous code worked fine AFAICT.

[l29ah]

Yeah, i agree that is a saner solution. @Brambor why the clamp?

[Brambor]

My bad. It would work with int, I believe. I made a mistake. As they say "use int instead of unsigned in CDDA codebase, you will avoid maintance errors". I made that error that day.

I see, it is not unsigned -1, but wrong order of arguments...

Yeah, revert it. I only made it to be more readable. But it is wrong now, as you report.

Sorry for the trouble I caused by "refactoring".

[l29ah]

force-pushed the reverting commit