[RFC] ASLA design, depth of code review and lack thereof

[viro@parcelfarce.linux.theplanet.co.uk]

Ladies and gentlemen, may I politely ask what description would fit somebody
who have made the following

        case SNDRV_PCM_FORMAT_FLOAT_BE:
        {
                union {
                        float f;
                        u_int32_t i;
                } u;
                u.f = 0.0;
#ifdef SNDRV_LITTLE_ENDIAN
                return bswap_32(u.i);
#else
                return u.i;
#endif
        }
and quite a few similar, er, wonders an ioctl?

That's right.  This code just has to be in the kernel.  It can't be in
a library, oh no.  It can't be a trivial macro that would result in
compiler generating the constant, no sir - it just had to be proudly
dumped into the great barfbag in the tree.

And that leads to a really interesting question: how many people had ever
read that code?  Or documentation covering that ioctl, while we are at it.

Unless I'm mistaken, ALSA used revision control for a long, long time.
Jaroslav, could you please find the origin of that little wonder and
share with the class - who had done that, why it had been committed into
ALSA tree and how did it manage to survive until the merge into the main
tree?