I have downloaded 1.8.5 and see small differences in the output of a simple sketch.
The reason for the differences I see are the filler used in _DBL. This is not set to anything so it takes over whatever the value the stack/heap has at that moment. However the working of the code in my test sketch is identical so far. Still I think there need to be a fix to zero that filler.
void float2DoublePacked(float number, byte* bar, int byteOrder=LSBFIRST)
{
_FLOATCONV fl;
fl.f = number;
_DBLCONV dbl;
dbl.p.filler = 0;
dbl.p.s = fl.p.s;
dbl.p.e = fl.p.e-127 +1023; // exponent adjust
dbl.p.m = fl.p.m;
#ifdef IEEE754_ENABLE_MSB
if (byteOrder == LSBFIRST)
{
#endif
for (int i=0; i<8; i++)
{
bar[i] = dbl.b[i];
}
#ifdef IEEE754_ENABLE_MSB
}
else
{
for (int i=0; i<8; i++)
{
bar[i] = dbl.b[7-i];
}
}
#endif
to be continued