Bug in struct with float arguments?

Started by AVC, January 11, 2022, 02:18:15 PM

Previous topic - Next topic



I love newLISP and I'm trying to use it for some graphics programming. So I'm writing a FFI module for the raylib library.

I've encountered an odd behavior of struct: If I define a 'struct' with one or more floats and pass it to 'pack', the float values are stored as zeroes. But surprisingly the same struct works well with 'unpack'. Here is an example:

(struct 'Vector2 "float" "float")

(pack Vector2 200 200) => returns "0000000000000000", which is wrong!

(pack "ff" 200 200) => returns "0000HC0000HC", which is what 'pack Vector2...' also should return

(unpack Vector2 (pack "ff" 200 200)) => returns (200 200) correctly!

I'm using newLISP 10.7.5 64-bit on Arch Linux with ffi and UTF-8 support enabled.

Is there a bug or am I doing something wrong?

Thanks in advance



Yes it's a bug in the handling of the "incoming cell data" for an ffi_type_float at nl-import.c:814.

That current line is
floatV = (float) *(double *)&cell->aux;
and should be replaced with the following
floatV = (float) ((cell->type == CELL_FLOAT)? *(double *)&uint64V : uint64V);
which corresponds to how it's handled in the pack function.


I've changed that line and recompiled. Bug solved! Thank you very much!