On Feb 6, 2011, at 8:57 AM, Byron Jeff wrote:

> Microchip just needs to be happy that a group hasn't done a =20
> purchase, requested and received the source, then redistributed it =20
> for free yet.

It would be technically possible for microchip to have implemented =20
their proprietary pieces in a way that does not require =20
redistribution, right?  Minimally, libraries and include files (for =20
user programs, not for gcc itself) are separate entities can can have =20
separate licenses?  The whole "register definition file" question is =20
particularly interesting in the microcontroller realm.  Not =20
technically part of the compiler, not something I want to create or =20
maintain myself, not something that is so stable (for a range of =20
micros) that you can do it once and be done.  MSPGCC (for TI MSP430) =20
apparently has recently switched to using TI-created .h files from =20
their own versions.  A good move, but on somewhat shaky ground WRT =20
"open source philosophy."


>>> [microchip charges $$$ for] not much more than repackaging and =20
>>> distributing it...

You left out "support."  A lot of "real" companies will happily pay =20
much higher costs to have a compiler with some amount of support =20
associated with it (hopefully a useful amount of support.)  Figure the =20
average small company spends a significant percentage (5% ?) of its =20
engineering time fiddling with "free" tools, and larger companies have =20
at least one FTE doing tool support (probably true even if the tools =20
are supported, but if paying $5000/y for external support cuts off a =20
significant percentage of this, there's no way it isn't a good deal.) =20
(interpreted uncharitably, the Gnu philosophy turns software authors =20
from artists into janitors.  Meh.)

BillW

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.