Better support and testing for empty string_array #246
Conversation
Signed-off-by: Scott K Logan <logans@cottsay.net>
|
@dirk-thomas, I'm re-requesting review because I added a change to handle some inconsistent behavior due to an assumption based on implementation-specific behavior in malloc. Here are some docs, which describe what might happen if you malloc/realloc with size 0: http://man7.org/linux/man-pages/man3/malloc.3.html |
The documentation for malloc states that an attempt to allocate memory of size 0 results in implementation-specific behavior. Some implementations return `NULL`, and others return a pointer that is expected to be `free`'d. Signed-off-by: Scott K Logan <logans@cottsay.net>
cottsay
force-pushed
the
empty_string_array
branch
from
958369a to
abed4ac
Compare
|
@cottsay do you want to try to get this and the other pr's you opened into rcutils before I do a 1.0.0 release? I can try to help make that happen, but I also do not want to delay too long. |
At this point, it would just be "nice to have." I'm not sure I'll have time to actually take advantage of these changes before the release anyway. This one and #252 are a higher priority IMHO - they both improve the behavior of existing functions. |
Since I closed #252, I think you should move ahead with the release without the other two API additions. Thanks, @wjwwood! |
|
I'm rerunning CI to include more, in case this change in behavior of string array causes a failure downstream:
|
|
Oops missed the fact that you merged it. I'll cancel those CI and see what the nightlys bring. |
Specifically test for
string_arrays of size 0, and relax the null check inrcutils_string_array_cmpwhen it won't be used.