3 ms·
I'm 35. I started learning SQL, C, C++, memory management etc. many years ago when I was a teenager. However, and I know that people will take the opportunity
by ilaksh 13y ago
I'm 35. I started learning SQL, C, C++, memory management etc. many years ago when I was a teenager.
However, and I know that people will take the opportunity to disrespect me for this, but I realized, after playing around with 3d C++ graphics in college, that I wanted to build interactive desktop applications quickly. So I learned to use a very powerful tool called Visual Basic 6. Not because I didn't understand memory management or anything else, but because it was a better software engineering decision for most projects since it allowed me to not repeat solving the same problems that had already been solved.
I know people aren't going to be able to understand this because of the reputation that Visual Basic has, and they will lose respect for me, but I am going to go ahead and say it anyway. Visual Basic 6 was and still is a better software engineering tool than C++. Why? One of the core principles of software engineering is DRY. Manual memory management means you are repeating yourself. It also means you are doing the memory management yourself rather than letting the computer do it for you. Another reason why Visual Basic 6 is superior software engineering over C or C++ is the component models.
I eventually moved on to .NET and C# because of the stigma of VB and also because the new frameworks provided better tools.
Eventually I realized that the closed source world of desktop Microsoft development was outdated since it was tied to one platform and old-fashioned business models. So I moved into web development. The most practical tool for web development was PHP. It was the best software engineering decision to select PHP to build a web application. Because it allowed me to avoid solving the same problems that many other engineers had already been faced with and solved on their own. Something that many software engineers don't understand is that not only is it important not to repeat yourself in your own code, but not to repeat solving the same problems other people have already solved for you.
So then I wanted to build a realtime interactive collaborative software application in the web browser. Because that is the most compelling, useful, and challenging type of application. So I tried to do it with PHP. Very awkward. I tried to use another popular server-side tool, Python. Again it turned out to be very awkward to do realtime collaborative software applications with Python (using things like Twisted).
So when I heard about this thing called Node.js that was built from the ground up to take full advantage of the single-threaded but asynchronous programming model of JavaScript as well as powerful underlying evented non-blocking I/O C libraries, I had to try it. Guess what? All of that software engineering done by the people who wrote those C libraries and the people who put together JavaScript and Node means that there are a lot of problems that I don't have to solve to write performant IO-intensive code on the back end! Continuation-based programming in JavaScript and Node.js just works better and is more maintainable (less code and less complexity) than the other solutions.
So I'm sorry, but if you're not adopting new programming languages, tools and frameworks, then I don't think you are a good software engineer. Because good software engineers only solve the problems that they have to, and don't continue to solve the same problems over and over.
Now -- NoSQL. So, it turns out that NONE of the many software applications I built over the years using Oracle, SQL Server or MySQL actually required 'real' db-level transactions or even involved very large datasets. And I believe that is the case for most programmers (not all of course). These days that is still true. Also, these days I only code in JavaScript (mostly CoffeeScript really). So when I heard about things like MongoDB that allowed me to store JSON directly, skip the object relational layer, and not even have to define a schema ahead of time, I had to seriously consider whether I needed a normalized relational database, SQL, to constantly
reiterate joins, etc. When you examine it logically, it turns out that we were actually doing all of those things just because THAT WAS HOW YOU DID IT. With the current ecosystem of JavaScript and JSON, it is BAD SOFTWARE ENGINEERING for the MAJORITY of software projects to use a relational database, joins, etc. Why is it bad software engineering? Because you are creating a bunch of problems for yourself that you wouldn't have if you used a different set of tools.
- timr 13y ago"So I'm sorry, but if you're not adopting new programming languages, tools and frameworks, then I don't think you are a good software engineer. Because good software engineers only solve the problems that they have to, and don't continue to solve the same problems over and over." Yes, and? You're arguing a straw man. I'm not claiming that things haven't changed for the better in 30+ years of coding, or that people should be writing web apps in FORTRAN, or that you shouldn't re-use code. But you're being...imaginative...if you believe that the latest shiny webapp tool (e.g. Node.js) is actually as new and innovative as you're claiming it to be. Anyone who has been around long enough to have written early Windows or Mac apps will have a lot of experience with single-threaded, evented programming. (Certainly enough to know that Node.js is a wonky regurgitation of a very old idea.) Same thing goes for your MongoDB apologetics: actually, no, the industry didn't settle on relational databases because "that was how you did it". They became widely adopted because they hit a sweet spot in terms of speed, reliability and flexibility (a spot that basically none of the current NoSQL products have yet to match). In other words, even if you like new things, experience helps you know when the new thing you're evaluating is a pile of crap. And judicious decision-making is far more important than the latest shiny when it comes to being a good engineer.
- ilaksh 13y agoYes, evented programming isn't new. Yes, and? So in your opinion Node.js and MongoDB are piles of crap? Then, just curious, what system would you use to build the back end for a realtime collaborative web application, or an API that needs to hit several web APIs and do other IO to service each request? And how do you recommend that we handle data storage for small business web applications? Really would like to know if there is something that is so much better than Mongo. You really think I should go back to building out fully normalized relational databases for small business web applications? What language/platform do you use, and how do you handle things like ORM? I would like to know if you have ideas that are better. In my experience, Node.js and MongoDB (throw CoffeeScript in there too) are not crap, perform very well, reduce lines of code, and are better software engineering decisions in most cases. Have you actually tried building applications that way?