[asterisk-dev] Adding "pipe" support to test infrastructure

[Philip Prindeville]

Hi,

I'm working on some changes to various parts of Asterisk that currently don't have good test coverage.

The easiest way (for what I'm doing) to add coverage is to interface with external commands that are known to provide correct results and test against them.

I've written a mockup of a function that runs a pipeline, collects stdout and stderr into buffers, and returns the processes exit code.

This is useful if, for instance, you generate data and want to verify it computationally internally, but then compare results with an external reference command.

It's also useful for commands that generate temporal (or ephemeral) results that can't use "canned" data in their generation.

Here's some details:

typedef struct {
	char *outbuf;
	size_t outlen;
	char *errbuf;
	size_t errlen;
	int exitcode;
} ast_test_capture_cmd_t;

int ast_test_capture_cmd_run(ast_test_capture_cmd_t *results, const char *cmd);

ast_test_capture_cmd_free(ast_test_capture_cmd_t *results);

Note that commands are run with stdin set to /dev/null.

What do you all think about adding this functionality?

Thanks,

-Philip