pkg/lwip: Use IS_ACTIVE() and bugfix#14665
Conversation
The `IS_USED()` macro doesn't work if macros are defined to `(1)` instead of `1`
|
More philosophic usage question: |
I'd say yes, if we are checking for the configuration macros directly we should stick to But why do we have two checks? There are places in contrib code where we check for |
In general I'd say, they are equivalent, yes. If the |
|
Squash? |
|
Something broke 😕 LWIP_IPV4=1 LWIP_IPV6=1 make -C tests/lwip_sock_tcp/ flash testDetails``` make: Entering directory '/home/mlenders/Repositories/RIOT-OS/RIOT/tests/lwip_sock_tcp' Building application "tests_lwip_sock_tcp" for "native" with MCU "native"."make" -C /home/mlenders/Repositories/RIOT-OS/RIOT/pkg/lwip Help: Press s to start test, r to print it is ready
*** halted. native: exiting make: *** [/home/mlenders/Repositories/RIOT-OS/RIOT/tests/lwip_sock_tcp/../../Makefile.include:772: test] Error 1 |
|
Oh no, its also broken in master :-( |
Regression was introduced in 035acc2 |
|
I did not mean for you to revert that change ... the change is valid. After some debugging, it seems more there is actually a bug in the test. The line in question checks if listen fails with an RIOT/tests/lwip_sock_tcp/main.c Lines 1087 to 1095 in 7e4b3d0 035acc2 might just have uncovered that test error. |
|
May I squash? |
|
Yes please. |
This also fixes an issue detected by `tests/lwip_sock_udp` when both IPv4 and IPv6 are enabled.
|
See #14668 |
|
Thanks for the reviews! |
|
Can you backport this to the 2020.07-branch, please in case I decide to make a point release? |
|
Backport provided in #14671 |
Contribution description
IS_ACTIVE()over#if/#ifdefsock_udp_get_local()Testing procedure
LWIP_IPV4=1 LWIP_IPV6=1 make -C tests/lwip_sock_udp flash testshould no passIssues/PRs references
Split out of #14622