Fix(trace): fix expected values and unknown address access

[YuShuanHsieh]

Why

1. The expected values should be changed after the sorting operation.
2. do_sort will access unknown memory address if the queue is null.

How

1. Fix the expected values of trace-4, trace-5.
2. always return true if it is a null queue.

[jserv]

Don't put "`" character here. Simply use quotation mark.

[jserv]

Rewrite strcmp(argv[1], "NULL") == 0 into !strcmp(argv[1], "NULL") for consistency.

[jserv]

It is a bit weird to me regarding the use of sort NULL. In my understanding, sort should be as generic as possible. It can accept almost anything including NULL. I don't think it is vital to introduce extra argument for sort since sort always perform the operations within queue. What is your consideration?

[YuShuanHsieh]

You are right.
The reason why I choose to add a new argument is to minimize the effect of other test cases. I think it may always return true if it is a null queue, but I am not really sure about this idea because honestly I am not quite familiar with all test cases.

[jserv]

Don't place check here. Instead, let's move to Line 477:

if (q == NULL)
    report(3, "Warning: Calling sort on null queue");
error_check();

The function error_check() would collect test information and interact with users.

[YuShuanHsieh]

Sorry, I’m not sure what you mean.
Cause in my point of view error_check()draw a scope of tests. like

error_check(); //reset error_occurred
// test case
return ok && !error_check();

So I still need to handle the null queue after q_sort(q) to prevent q->head bad access.
Could you give me more about your opinion? Thx :)

[jserv]

You can follow the similar flow in function do_size and do_reverse, that just report when NULL pointer is inputed. qtest itself does not need to take care about fatal condition. Instead, qtest only reports to users and each trace file denotes the criteria of test scenario. In this case, do_sort attempts to report the checkpoints without managing to recover.

Consequently, the implementation behind queue would have to couple with the above though.

[YuShuanHsieh]

I probably catch your point. Please review it again. Thanks.

[jserv]

Thank @YuShuanHsieh for contributing to this project!