News for hackers. Stuff that matters.
Re:What senior engineers do: fix knowledge holes by z3t4 on Saturday July 20, 2019 @05:45PM (#20488573)
Could start with regression testing, each time you break something, write a test so that you do not break the same thing again. After a time you will have a decent test suit. Also write tests for all new stuff. And also write tests for everything you change. Would also be an idea to start using version control, so that each change is documented in the commit message. Both tests and SCM you can start with on any stage eg. even on legacy code. Another strategy would be to move some functionality into a micro service. For example instead implement new feature X, make it a standalone service that not only is decoupled from the monolith, it should not even share the same db, it should be completely standalone.
Re:What senior engineers do: fix knowledge holes by EliRivers on Saturday July 20, 2019 @06:04PM (#20488649)
While we do use some of those methods, sadly, we are not in possession of much of the hardware that our software runs. Making it impossible for us to fully test ourselves; we can usually bank on a period of testing on a customer site during installation, but after that, we're out of luck. Often the hardware claims to support some protocol, but we've found that really, that means "a variation on the theme of a given protocol" so ensuring that we still meet the protocol isn't quite a guarantee of success. Customers are often sympathetic to the idea that their hardware doesn't quite work as advertised - but not very sympathetic.

There is often literally no way for us to know that we've broken it until a customer upgrades and discovers it no longer runs their hardware as it used to. Which does happen.

Re:What senior engineers do: fix knowledge holes by elcritch on Saturday July 20, 2019 @11:46PM (#20489855)
Can you record the actual machine protocol? Then use that recorded data in regression/integration future tests.