In order to ensure that libsir is working properly on your system, the first thing you should do after compilation is verify that the test suite runs successfully (in fact, we recommend making it part of your deployment workflow).

It will exit with 0 if all test(s) ran successfully and passed, or 1 if errors occurred/any test failed.

Basic usage

The simplest way to achieve this is simply to run build/bin/sirtests from the root of the repository. You should see output similar to the following:

done: all 32 libsir tests passed in 5.085sec!

Advanced usage

sirtests has a command-line interface that allows you to control its behavior. For the current options, run sirtests --help. You should see output similar to the following:

Usage:
 
    --perf       Only run the performance measurement test
    --only       name [, name, ...] Only run the test(s) specified
    --list       Prints a list of available test names for use with '--only'
    --leave-logs Log files are not deleted so that they may be examined
    --wait       After running test(s), waits for a keypress before exiting
    --version    Prints the version of libsir that the test suite was built with
    --help       Shows this message

Of note here is the obvious one, --perf. The perf test is only run if you explicitly specify this flag. It is a good way to ensure that libsir is compiled correctly for deployment in a production environment. If you get a very slow (or dubiously fast) result, you should re-examine your configuration.

The result includes a comparison to an equivalent raw printf call for the same number of iterations. I expect this to always be faster than libsir for many reasons, but I have seen it beaten once.

Here is an example of the results of --perf on this development machine (iMac Pro, 3.2GHz 8-core Xeon, 64GB 2666MHz DDR4, SSD):

printf: 1000000 lines in 3.375sec (296296.3 lines/sec)
libsir (stdout): 1000000 lines in 5.427sec (184263.9 lines/sec)
libsir (file): 1000000 lines in 1.278sec (782472.6 lines/sec)
timer resolution: ~1000nsec

Remarks: The perf test only outputs to the debug level. If level switching were introduced where formatting options varied from level to level, a much slower elapsed time could be expected, since some of libsir's internal formatting buffers would need to be recalculated each time.

The other useful flags include --list and --only if you wish to narrow down a problem test or set of tests. Please let us know if you think of additional tests that should be performed by opening a feature request.