Add in some new .in files, but still keep original files in distribution
files.
Essentially, use coap@LIBCOAP_API_VERSION@ (or equivalent) instead of
coap2. so resultant files have the correct value.
coap_config.h.lwip.in: (new)
coap_config.h.riot.in: (new)
include/coap2/coap.h.windows.in: (new)
As per https://gcc.gnu.org/onlinedocs/cpp/Search-Path.html, many of the
coap2/ prefixes are not required for #include, so these have been removed.
Make sure that the test suite accesses the latest source version of
coap@LIBCOAP_API_VERSION@/coap.h by creating a tests/test_common.h.in that
makes the correct tests/test_common.h which is included in all of the test
suite files.
src/coap_io.c:
ifdef out IPv4 struct in_pktinfo code that does not compile, copying what
has been done later in the same coap_network_send() function.
tests/test_session.c
Correctly reverse the parameters to coap_new_client_session() in t_session6()
so that the test application does not bind to local-address, port 5683 twice.
[It seems that other distrubutions are happy with this when just SO_REUSEADDR
is used on the socket, but FreeBSD needs SO_REUSEPORT enabled as well.]
BUILDING
Record that gmake is required to build TinyDTLS for FreeBSD, not make.
This change replaces the call of abs(v - ref) to avoid a compiler
warning for -Wabsolute-value:
test_session.c:95:11: warning: taking the absolute value of unsigned type 'unsigned int' has no effect
[-Wabsolute-value]
return (abs(v - ref) / 100.0) <= FP_ERR_THRESHOLD;
Created new 'CU_ASSERT_PTR_EQUAL_C(actual, expected)' to handle const variable matching
Where appropriate
Replace 'unsigned char' with 'const uint8_t'
Replace 'str' with 'coap_str_c_t'
Replace 'str' with 'coap_string_t'
Prefix with 'const' some type casts
Update coap_add_attr() calls inline with updated API
Update coap_resource_init() calls inline with updated API
Make use of coap_make_str_const() for defining parameters
tests*/.c
Update code to stop memory leaks (as found by valgrind)
examples/coap-rd.c
Track the fact that uri_path in coap_resource_t is now a pointer
Jon Shallow