So in the whole area of code analysis, this seeks to discover syntactic errors, logical flaws, and other weaknesses and vulnerabilities that may be embedded in the code itself. So overall, code analysis should be done at every level: unit, stream, subsystem, full system, et cetera, since the interactions become more complex and variable and can produce different kinds of results with changes.

Now, the process will, of course, involve various techniques, some of which will be automated, some manual, some will be static while others will be processed as dynamic. Now, automated code review software checks source code for compliance with a predefined set of rules or best practices. Manual code reviews are processed on small portions of code by programmers, not the builders, it should be pointed out, to visually examine the code and its flow.

Static methods are combinations of the above, looking at the code in different ways but without actual execution. Whereas dynamic methods execute the code or portions of it in differing context to observe actual performance. And this would include thread management, process isolation, error handling, and so forth. Now, one of the things we need to examine first before going further is that there are weaknesses of source code analysis tools. It can be that many security vulnerabilities are difficult to find automatically, simply because they may not be something that the program tool were using its program to find exactly what we're looking for.

Every automated tool suffers from the plague of potential false positives and negatives. And this environment would be no different. We frequently cannot recognize configuration issues through automated source code analysis. It can be difficult to prove that a thing that we believe is a vulnerability in the absolute sense is not in fact a vulnerability in the context being examined. And the difficulty of analyzing code that cannot be compiled leads to other linguistic and syntactic problems.