Act Contractor Forms: Fill & Download for Free

What’s the biggest no-no when you are a software engineer? I’ve been writing software for 50 years. Here’s my response.The biggest no-no that you can make as a software engineer is to fail to test your code.Print the following statement out and paste it where you can see it when you write or modify code:Untested code doesn’t work.No matter how much experience you have, or how much confidence you have in your skills, the simple fact is that errors can, and will, arise from a variety of circumstances. You could have misread the specification or misunderstood the requirements. You could have miscoded the solution in a manner that compiles and executes but fails to accomplish your desired outcome. Your code or changes might be fine by themselves but trigger a heretofore unknown downstream problem. Your changes might handle low-volume transactions just fine but prove to be too costly when placed under the high volume of data present in an online system. The compiler, interpreter, runtime libraries, or other dependencies might have bugs that were unknown until you made your changes. The version of the operating system that you used to create and test your work might be fine but the customer, still on an older release of the OS, might run into issues.Code walkthroughs, inspections, peer review, pair programming, agile methods, and so on, are all great techniques for improving the overall quality and productivity of a team. But they are not a substitute for subjecting your code to a comprehensive set of unit tests and system-level regression tests…in a test environment that closely matches the production environment.Remember the initial failure of the Affordable Care Act web site under its first real world load? While I wasn’t at all involved in that effort, it was pretty clear to me that the contractor(s) had not taken the time to conduct any worthwhile performance tests. (My memory is that it could handle only a few transactions per minute; I don’t recall the exact details. Perhaps someone who reads this and does remember the details can add them here in a comment).There are lots of ways that a software engineer can mess up. I just think that failing to test your code is the worst failure we can make, in part because it is in a class of errors that is unique to software engineering.My customers would often complain that they could not afford a test environment that had similar performance and complexity characteristics to their production environment. I would point out that they needed to look at the challenge like a software engineer. Simulate the stuff you can’t afford to duplicate. You don’t need 18 test links to 18 different credit authorization networks; you need 18 simulated links and simulated data. What with the GDPR and other privacy laws, you don’t want to use real data anyway. Besides, in a simulated environment, you can re-run a test over and over using the same data stream, which is really helpful when you are chasing a Heisenbug-type problem.Test your code. Always.

Just another answer to augment the rest.I have spent around 15+ years working as an independent contractor for various Silicon Valley high tech (often pre IPO) companies in leading edge software development R&D. The following comments only apply to this narrow but critical niche.The most critical issue is that client company and contractor maintain a flexible professional relationship that ultimately puts the clients success first and foremost. R&D by definition is unpredictable and malleable requiring flexibility and innovative adaption. Again the client's success is mutually beneficial for both the client and the contractor as the client needs to generate sales / revenue and the contractor needs to maintain a stellar reference history of known successful clients. Success begets success.One can over analyze and over spec anything as you cannot squeeze blood from a turnip. One cannot focus on procedural measures but in the end these will not assure success as R&D is a creative process. People solely focused on process are the fodder of Dilbert cartoons.The key requirements are the proficiency of both the client and the contractor(s) and that their long term interests align. The ideal scenario is when the contractor(s) behaves and acts like client employee(s).A disorgenized client will typically fail, a contractor solely motivated by money will typically fail.

When outsourcing your work, it’s always important to use a company where you can find reputable contractors, and where the company can act as mediators if there are any questions or concerns.Make sure you thoroughly interview your contractor(s), asking them about their previous work experiences, their work hours, your expectations of communication, the deliverables and deadlines. Put all these parameters in writing and have both parties sign it.Keep in mind your budget before hiring anyone and be clear as possible about this budget. Don’t be afraid to check in on progress as well (without micromanaging). Taking the right steps before outsourcing will help ensure that you get the work done to your expectations.Hope that helps, and good luck!