Re: [PATCH] tests: unit: simplify test-visitor-serialization list tests

[Stefan Weil]

Am 05.09.22 um 13:00 schrieb Paolo Bonzini:
> test-visitor-serialization list tests is using an "if" to pick either the first
> element of the list or the next one.  This was done presumably to mimic the
> code that creates the list, which has to fill in either the head pointer
> or the next pointer of the last element.  However, the code in the insert
> phase is a pretty standard singly-linked list insertion, while the one
> in the visit phase looks weird and even looks at the first item twice:
> this is confusing because the test puts in 32 items and finishes with
> an assertion that i == 33.
>
> So, move the "else" step in a separate switch statement, and change
> the do...while loop to a while, because cur_head has already been
> initialized beforehand.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>   tests/unit/test-visitor-serialization.c | 157 +++++++++++-------------
>   1 file changed, 69 insertions(+), 88 deletions(-)
>
> diff --git a/tests/unit/test-visitor-serialization.c 
> b/tests/unit/test-visitor-serialization.c
> index 907263d030..667e8fed82 100644
> --- a/tests/unit/test-visitor-serialization.c
> +++ b/tests/unit/test-visitor-serialization.c
> @@ -427,131 +427,117 @@ static void test_primitive_lists(gconstpointer opaque)
>       ops->deserialize((void **)&pl_copy_ptr, serialize_data,
>                        visit_primitive_list, &error_abort);
>   
> -    i = 0;
> +
> +    switch (pl_copy.type) {
[...]> +    default:
> +        g_assert_not_reached();
> +    }
>   
>       /* compare our deserialized list of primitives to the original */
> -    do {
> +    i = 0;
> +    while (cur_head) {
>           switch (pl_copy.type) {
>           case PTYPE_STRING: {
[...]
> @@ -578,9 +559,9 @@ static void test_primitive_lists(gconstpointer opaque)
>               g_assert_not_reached();

As both switch statements have the same 12 cases plus a default case 
with g_assert_not_reached(), a static code analyzer might complain that 
the 2nd default case will indeed never be reached because the first one 
already raises an assertion. So the code in the 2nd default case could 
be removed.

Regards,
Stefan

>           }
>           i++;
> -    } while (cur_head);
> +    }

OpenPGP_0xE08C21D5677450AD.asc
Description: OpenPGP public key

OpenPGP_signature
Description: OpenPGP digital signature