[systemd-devel] [PATCH 0/3] Using assert_se() on actions with side effects on test cases

[Lennart Poettering]

On Mon, 25.08.14 22:05, Filipe Brandenburger (filbranden at google.com) wrote:

Applied all three! Thanks!

And yes, I agree, all tests should use assert_se(), never
assert(). Quite a few tests snuck in though that used assert()
instead. WOuld be happy to take a patch correcting this!

> I bumped into these when building systemd with CPPFLAGS='-DNDEBUG' (which is
> the default for one of my build environments) which ends up optimizing out the
> assert() statements. It turns out that they were being used in some places in
> test cases where there was a side effect, so optimizing them out would cause
> the tests to crash.
> 
> More to the point, maybe it would make sense to have *all* the assertions in
> tests be assert_se. Maybe a global search/replace?
> 
> Not sure if this issue is not present elsewhere in systemd either... Not sure
> if it's really worth supporting -DNDEBUG. Considering systemd is already
> redefining assert(), maybe make it unconditional?
> 
> Cheers,
> Filipe
> 
> 
> Filipe Brandenburger (3):
>   test-compress: make sure asserts with side effects use assert_se()
>   test-path-util: use assert_se in all assertions
>   test-util: use assert_se() for call to safe_mkdir with side effect
> 
>  src/journal/test-compress.c |  4 ++--
>  src/test/test-path-util.c   | 30 +++++++++++++++---------------
>  src/test/test-util.c        |  2 +-
>  3 files changed, 18 insertions(+), 18 deletions(-)
> 


Lennart

-- 
Lennart Poettering, Red Hat