If all the source that compiles to ARM code is now open, that means that
alternatives to GNU/Linux could be ported to run on the rasPi: Plan 9, a BSD
port, RISC OS, Mach, even some experimental home-brewed system. The educational
value to the budding OS architect or the demoscene programmer should be quite
apparent.
The educational value of a modified graphics DSP software, on the other hand, is
pretty low, given the intended users and usage. I see only two present, actual
justifications for replacing it, which pointedly excludes speculation such as
bugfixes, added features and performance improvements. One is to bypass
Broadcom's feature license management for other contributors' technology (e.g.
MPEG LA, Macrovision, Microsoft), which would negatively impact Broadcom's
relationships with those contributors as well as Broadcom's revenue model, and
thus the RP Foundation's ability to offer the rasPi at its famously low price
point. The second is to add acceleration for new video codecs, which is
legitimate and desirable, but probably impractical. While Broadcom might be
willing to open up info on standard peripherals, I strongly doubt they would do
the same for their apparently novel "2-dimensional" (or is it just
vector?) graphics DSP core, much less distribute a free SDK for it, especially
if they're on the hook to Keil or CodeSourcery or other compiler vendor for
per-seat fees.[ Reply to This | Parent | # ]
|