Done right there will be a lot more in there than syntax errors. A good example of this sort of test will separate those who have revised syntax without really thinking (they'll get the syntax errors but little or nothing else, great if you want to employ a human linter) from those who actually think about what they are looking at will spot much more (a logical error that would cause an infinite loop, a point where a comment and the code disagree, an incorrectly validated input, an injection flaw, a heavy expression inside a loop that could be done first and result-cached for efficiency (your compiler cannot always identify such situations), hard-coded credentials, bad naming, ...). The best may identify some but also state why in the grand scheme of things is might not matter (efficiency in code that runs once so 10μs and 100μs is statistical noise) or where other priorities might take precedence (for example readability and therefore ease of maintenance).