Re: [Tarantool-patches] [PATCH luajit 1/2] test: introduce asserts assert_str{_not}_equal

[Sergey Kaplun via Tarantool-patches <tarantool-patches@dev.tarantool.org>]

Hi, Sergey!
Thanks for the patch!
Please, consider my comments below.

On 31.10.23, Sergey Bronnikov wrote:
> From: Sergey Bronnikov <sergeyb@tarantool.org>
> 
> The patch follows up commit a0483bd214f2 ("test: introduce module for C
> tests") and adds additional asserts suitable for comparing strings.
> ---
>  test/tarantool-c-tests/README.md |  2 ++
>  test/tarantool-c-tests/test.h    | 17 +++++++++++++++--
>  2 files changed, 17 insertions(+), 2 deletions(-)
> 
> diff --git a/test/tarantool-c-tests/README.md b/test/tarantool-c-tests/README.md

<snipped>

> diff --git a/test/tarantool-c-tests/test.h b/test/tarantool-c-tests/test.h
> index 8b14c705..bbf573b2 100644
> --- a/test/tarantool-c-tests/test.h
> +++ b/test/tarantool-c-tests/test.h

<snipped>

> @@ -214,4 +212,19 @@ static inline int todo(const char *reason)
>  	);								\
>  } while (0)
>  
> +#define assert_str_equal(got, expected, n) do {				\
> +	assert_general(strncmp(got, expected, n) == 0,			\
> +		       ASSERT_EQUAL_FMT(int, "%d"),			\

Typo: s/int/str/

> +		       __FILE__, __LINE__, (got), (expected)		\
> +	);								\
> +} while (0)
> +
> +#define assert_str_not_equal(got, unexpected, n) do {			\
> +	assert_general(strncmp(got, expected, n) != 0,			\
> +		       ASSERT_NOT_EQUAL_FMT(int, "%d"),			\

Typo: s/int/str/

> +		       __FILE__, __LINE__, (got), (unexpected)		\
> +	);								\
> +} while (0)

If some test fails we got the following output:

| TAP version 13
| 1..1
| not ok 1 - lua_concat_testcase
|   ---
|   location:     /home/burii/reviews/luajit/lj-881-lua-concat/test/tarantool-c-tests/lj-881-fix-lua-concat.test.c:81
|   failed_assertion:     assert_str_equal
|   got: 1652880104
|   expected: -1138225337
|   ...
| # Failed 1 tests out of 1

Failed assertion field is incorrect (see the comment above).
But the most important is "got" and "expected" fields, that returns the
addresses of strings, which isn't very meaningful.

I suggest dumping the strings instead if they are not long enough (less
than 128 symbols, I suppose). The maxdump string length should be
a customizeable parameter. I suggest defining a macro `MAX_DUMP_STRLEN`
inside the <test.h> header. So the user can redefine it before the
`assert_str_{not_}equal()` and use a custom value.

If the string is too long, we should dump the offset of the mismatched
symbol. Or maybe it's better to always dump it.

Thoughts?

Side note:

Also, this comparing "by eye" isn't very convenient if values aren't
aligned, so maybe it is better to use spaces instead of tabs to align
values? This may be added within the separate patch series later.

> +
> +

Nit: Excess empty line.

>  #endif /* TARANTOOL_LUAJIT_TEST_H */
> -- 
> 2.34.1
> 

-- 
Best regards,
Sergey Kaplun