Re: [Tarantool-patches] [tarantool-patches] [PATCH v2 2/2] say: take getaddrinfo() errors into account

[Alexander Turenko <alexander.turenko@tarantool.org>]

> @@ -72,14 +72,28 @@ static void
>  test_getaddrinfo(void)
>  {
>  	header();
> -	plan(1);
> +	plan(3);
>  	const char *host = "127.0.0.1";
>  	const char *port = "3333";
>  	struct addrinfo *i;
>  	/* NULL hints should work. It is a standard. */
>  	int rc = coio_getaddrinfo(host, port, NULL, &i, 1);
>  	is(rc, 0, "getaddrinfo");
> -	freeaddrinfo(i);

Why do you remove freeing of the previous result? It will lead to a leak
and so will require extra work to run unit tests under ASAN or Valgrind.

> +
> +	/* gh-4138: Check getaddrinfo() retval. */

Nit: retval and diagnostics area. You changed the code and the comment
becomes inaccurate.

> +	rc = coio_getaddrinfo("non_exists_hostname", port, NULL, &i,
> +			      15768000000);
> +	isnt(rc, 0, "getaddrinfo retval");
> +	const char *errmsg = diag_get()->last->errmsg;
> +	const char *exp_errmsg_1 = "getaddrinfo: nodename nor servname provided"
> +		", or not known";
> +	const char *exp_errmsg_2 = "getaddrinfo: Servname not supported for "
> +		"ai_socktype";
> +	const char *exp_errmsg_3 = "getaddrinfo: Name or service not known";
> +	bool is_match_with_exp = strcmp(errmsg, exp_errmsg_1) == 0 ||
> +		strcmp(errmsg, exp_errmsg_2) == 0 ||
> +		strcmp(errmsg, exp_errmsg_3) == 0;
> +	is(is_match_with_exp, true, "getaddrinfo error message");
>  
>  	/*
>  	 * gh-4209: 0 timeout should not be a special value and
> 
> > The patch looks okay other then this.
> 
> commit d9c23fde4c7c1fa2979954be83c1a2f6f8407347
> Author: Roman Khabibov <roman.habibov@tarantool.org>
> Date:   Tue Jul 30 15:39:21 2019 +0300
> 
>     coio/say: fix getaddrinfo error handling on macOS
>     
>     Before this patch, branch when getaddrinfo() returns error codes
>     couldn't be reached on macOS, because they are greater than 0 on
>     macOS (assumption "rc < 0" in commit ea1da04 is incorrect for
>     macOS).
>     
>     Note: diag_log() in say.c was added, because otherwise it will be
>     hid by the following diagnostic and then say that it should
>     be handler in a better way after #1148. Also, two diag_set() in

'and then say' -- looks as copy-paste from review notes.

Typo: handler -> handled.

>     syslog_connect_unix() was added to avoid asserts in this
>     diag_log().
>     
>     Need for #4138