signed value raw to physical error

[Former user]

If a signal is less than 64 bit width and it's physical value is negative，then calculate physical value just by

double Signal::rawToPhysicalValue(double rawValue)
{
    /* physicalValue = rawValue * factor + offset */
    return rawValue * factor + offset;
}

will result in a hugn value, which is not the real physical value.

The singed bits '1' filled in function

uint64_t Signal::decode(std::vector<uint8_t> & data)

are all recognized as valid value bits.

Hope reply，thank you！

[Tobias Lorenz]

I'll check it on the weekend.

Currently I don't see the issue in the library, although the data types are different.

So filling in signed 1 bits for a negative value is correct for a uint64_t in decode, right? And doing calculations of positive/negative values with a double also seems correct in rawToPhysicalValue.

I guess, the issue is the cast in between from uint64_t to double. So how do you do this? Is the rawValue as double correct, after the cast?

[Rock Shawn]

I wrote my own rawToPhysicalValue function, in which i did the mandatory type conversion to rawValue according to the signal size and signed type before

return rawValue * factor + offset;

That's may not a nice solution, but i got the correct physical Value by doing it.

[Tobias Lorenz]

Ok, so you basically followed my suggestion. You wrote an own function that does signal size and signed type casts before passing it to rawToPhysicalValue?

[Tobias Lorenz]

So, is this still still a major bug, or can I downgrade this?

[Rock Shawn]

Yes, you can downgrade it. I only want to report my trouble in using this library. So, is that indeed a problem, or i am wrong using it ?

Thanks！

[Tobias Lorenz]

Yes, maybe it's wise to add some examples on how to do the type conversions properly.

Also the unit test (test_Signal.cpp) only shows the functions separately: decode/encode and rawToPhysicalValue/physicalToRawValue.

[Tobias Lorenz]

I extended the test_Signal.cpp to show how decode and rawToPhysicalValue works in combination. I used a union instead of an explicit cast to access the raw data with the right signedness. I hope that gives an appropriate example.

[Tobias Lorenz]

Resolved with this commit: https://bitbucket.org/tobylorenz/vector_dbc/commits/969a7712251467a636b2be8e9b09a5f1793afeb2

[Rock Shawn]

Using a union is a better way, thank you!