assert() cannot be used in expression context

[sonydevworld]

assert() cannot be used in expression context

[kamtom480]

@dhalbert Thank you for letting me know. I'm currently checking if we can change the assert in NuttX to also work in CircuitPython. But this is not a quick solution. This will take time.

However, in the meantime I noticed that the Spresense SDK is built with the NDEBUG flag: https://github.com/sonydevworld/spresense-exported-sdk/blob/c12296c5ffbd0779e630a653c76a78558b463271/nuttx/.config#L58

But the configuration was not used in CircuitPython. Below is a patch that fixes this:

diff --git a/ports/cxd56/Makefile b/ports/cxd56/Makefile
index 238edb96e4..81b1b171e3 100644
--- a/ports/cxd56/Makefile
+++ b/ports/cxd56/Makefile
@@ -23,6 +23,7 @@
 # THE SOFTWARE.
 
 include ../../py/circuitpy_mkenv.mk
+include ./spresense-exported-sdk/nuttx/.config
 
 CROSS_COMPILE = arm-none-eabi-
 
@@ -91,6 +92,10 @@ CFLAGS += \
 	-fdata-sections \
 	-Wall \
 
+ifeq ($(CONFIG_NDEBUG),y)
+    CFLAGS += -DNDEBUG
+endif
+
 OPTIMIZATION_FLAGS ?= -O2 -fno-inline-functions
 
 # option to override compiler optimization level, set in boards/$(BOARD)/mpconfigboard.mk

At this point you should be able to build CircuitPython without any problem with Spresense. However, when the NDEBUG flag is enabled, assert will do nothing.

Maybe CircuitPython could have its own assert implementation? What do you think about this?

[dhalbert]

Thanks for your comments. Right now we simply removed the troublesome assert, and that may be all we need to do. In the long run perhaps we could talk to the nuttx people about the issue. It does seem to be a rare problem, though. We will also see about NDEBUG, as you mentioned.