You are right about posix compliance, this is a valuable goal, but at
the same time it raises the hard remark:
At some point NuttX will grow too large for deep embedded platforms.
Sebastien
On 3/8/23 21:41, Gregory Nutt wrote:
Historically, whenever we find a POSIX issue we have always fixed it
in order to as compliant as possible. In the hierarchy of values,
POSIX is probably at the top of the list and well above personal
preferences and novel solutions. In the name of POSIX compliance, we
have eliminated architecture support, increased code size, forced
redesigns, etc. So I think it is pretty hard to come with a truly
convincing argument why we should support a clearly non-compliant
behavior (or lack or behavior).
On 3/8/2023 2:35 PM, Nathan Hartman wrote:
On Wed, Mar 8, 2023 at 3:20 PM Gregory Nutt <spudan...@gmail.com> wrote:
POSIX defines the TERMIOS options and, I suspect that those TERMIOS
options are required, not optional (need to check that). If that is
true, then there should be no CONFIG_SERIAL_TERMIOS option; it should
always be enabled.
Unless the user (like me) knows termios will never be used. I have both
scenarios, used (serial config changes at runtime -- connected to
external
equipment) and never used (serial config set once at build time --
connected to a chip on the circuit board).
Even though we want as close to POSIX compliant as possible, we still
should allow to eliminate code to reduce flash footprint whenever
possible.
Otherwise we won't be for deeply embedded anymore.
Cheers
Nathan