Go Integer overflows translated from large C enum numbers about c-for-go HOT 5 OPEN

I understand the problem. At the same time, the statement

In C, some legit enum definitions

is kind of questionable, I'm afraid. According to C99, enum is supposed to be represented by an integer, however, it is not specified which actual type or signedness is to be used. The meaning of an enum value STATUS_FAILED = 0xFFFFFFFF therefore depends on the compiler implementation. This means it is not as "legit" as it may seem.

That being said, it would probably make sense to use e.g. gcc as reference. But this is not a trivial problem. As enums can be evaluated, they can overflow during evaluation, and if the expression is evaluated in Go (because it is not evaluated before translation), the enum value may have a different result.

from c-for-go.

google/flatbuffers#5161

not so easy even in C/C++ codebases

from c-for-go.

https://github.com/RadeonOpenCompute/rocm_smi_lib/blob/master/include/rocm_smi/rocm_smi.h#L130

Sometimes people do use this pattern in enums.

There are workarounds for it. So if it's too much effort for a corner case, I'm all for closing this ticket.

from c-for-go.

@hxy9243 definitely not closing, coz we may eventually find a workaround. The target of c-for-go is to be as practical as possible

from c-for-go.

Sure. Right now I'm using hard-coded bash and sed to translate certain numbers in C header files. Would love a more elegant solution with c-for-go.

from c-for-go.