Subject: Re: [linux-audio-dev] sfront 0.65 09/07/00
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Sat Sep 09 2000 - 21:45:12 EEST
>trick on float arrays, and I had thought IEEE floating-point
>didn't have machine-specific byte-order issues (maybe I'm wrong
>on that).
i used to think this too, till jaroslav on the alsa-devel mailing list
made me eat humble pie ... :)
IEEE floating point specifies the meaning of the bits, but not the
order of the bits within a multibyte extent, in somewhat of a similar
way to the way that a 2 bit integer has the same constant "meaning"
for each bit, but the order of the bits differ across chip
architectures. IEEE floats on a big-endian machine can (and are) have
different byte-level "appearance" than the same value on a
little-endian machine.
i was really baffled by this, but it has been demonstrated to me the
hard way that this really is true.
--p
This archive was generated by hypermail 2b28 : Sat Sep 09 2000 - 22:28:18 EEST