Whether you learned from a University, or mentor, or what have you (I'm mainly more concerned with University/Equivalent), what did the institution do right? And what do you think they could've improved upon when learning your programming skills?
I'm curious how everyone felt their institution did as far as teaching them how to become a good programmer.
I was not a comp sci major, but one comment I'll make: Undergrad education is supposed to be focused on timeless fundamentals of a field, not the latest buzzwords and technologies or mundane nitty-gritty practical details. If you want to learn the absolute latest research and buzzwords, that's what research journals and/or grad school are for. If you want to learn nitty-gritty practical stuff like source control and maintenance, that's what real-world experience is for.
I majored in biomedical engineering, and I didn't understand this at the time. I always wondered why we weren't learning about "latest cool buzzword that will get me a job X" and instead were wasting time on free body diagrams or reaction kinetics something boring like that. In hindsight the focus on the timeless fundamentals of engineering and biology in my undergrad education makes perfect sense. The "latest cool buzzword" changes too fast and is hard to understand deeply without a solid grasp of the fundamentals.
Teaching lots of detail about source control and the latest development methodologies (agile, waterfall, RAD, SCRUM, or whatever else people use) is silly because it will be obsolete in 5 years, there's nothing conceptually deep about it, and it's easy to learn on your own. The timeless fundamentals of computer science are computer architecture, algorithms, complexity classes, data structures, the Church-Turing thesis, etc.
Right: Teach us multiple levels of abstraction in programming languages - we studied Assembly, C, C++, and some Java. I believe that it's important for developers in high-level languages like Java to still understand what happens behind the scenes.
Wrong: No maintenance skills or exercises, we always wrote code from scratch.
Wrong: No test driven development, no agile methodology at all, too much "when I was your age writing punch card software" lecturing, no source control, no refactoring, no design patterns, etc.
Right: Free pizza at ACM meetings.
Pros: Our university had a great programming languages class that helped me start to really wrap my head around how programming languages work. I had a fantastic professor that taught the importance of design, best practices, and version control.
Cons: Lots of (very obvious) cheating, too many apathetic professors who didn't challenge us, especially in important classes like operating systems and assembly...they would pass anyone as long as they turned in anything.
To preface, I was a Information Systems major (Business College) at a public state university.
Right
Wrong
The two most important courses that I took as an undergraduate were computer organization and computer architecture. These courses were cross-listed with the EE department when I was in school and were taught from a computer engineering perspective. By the time a candidate completed this course sequence, he/she had a firm grasp of how computational devices worked all the way from half-adders through hierarchical memory systems. High-level software development is a piece of cake after one has implemented an instruction set in microcode.
With the above said, I believe that the number one problem that I have with most computer science programs today is that they have morphed into glorified vocational programs (what Joel Spolsky refers to as “Java Schools”). There is far too much emphasis on high-level coding and far too little emphasis on the design and application of computational automata. This orientation results in a temporal education that has a half-life of about five years.
Right:
Wrong:
Right:
Wrong:
I could go on and on. Maybe you can narrow down your question?
Right:
Wrong:
I did not expect my university to teach me languages beyond the introductory C++ course (which I already had, and got credited). What I expected was theory of computer programming, analysis, and software development. What I got was outdated theory of computer programming, analysis and software development.
Things I feel were done right within my university mostly in terms of formalizing things:
Side effects from my university education worth mentioning:
Areas for improvement:
Right:
Wrong:
At the end of it, my diploma program was really a taster of quite a few different things, some of which I pursued on my own time and after college. The downside was that I later learned there was much more to learn!
I live in Québec and I don't think our local college equivalents have the same targets American colleges have. Once you get out of CÉGEP (that's how we call them here), you should be ready to face the brave new world of finding a job. (Which, apparently, is not what most computer science students do. In my prom, there's only one person who did not continue to university.)
Still, knowing this, CÉGEP teachers tend to teach modern and immediately usable technologies, with a much smaller background on data structures and real computer science.
I'm not sure yet if this is good or not. However, our motivation for school projects all got us much, much farther than the curriculum alone would have. I'd say that the best things we had were teachers with a vision of their topic.
It's easy to vomit theory on your students and expect them to learn and remember it. My feeling however is that the classes where this was the case were the least productive ones (though maybe that's because I nearly never attended them). However, when there was a philosophy behind the class, and we were taught things because the teacher felt they were important (versus being taught X because the curriculum says we'll learn X), that was a whole new world of interesting things.
Therefore, I believe having good teachers is more important than anything else. If they didn't teach you about something, they'll probably have transferred you the reasons to teach it to yourself. That's what happened to source control, design patterns, and the .net framework: the program didn't cover any of these, but we used them all.
I'm currently an undergrad, but here are some things I'd like to improve on:
More cross-pollination between computer engineering and computer science programs. Right now CS students take the engineering digital logic course, and CE students take two semesters of Java, data structures, and two CS electives (software development and operating systems being common). I would love to take, say, an embedded systems course, but the list of prerequisites is prohibitively long because I'm not an engineering student.
More exposure to languages. The first two classes every CS major takes are Java (programming fundamentals and object-oriented programming), then architecture, then data structures in Java again. So, by the end of the fourth semester, a student has only been required to see Java and MIPS. I'm taking the programming languages course now, and this is the first exposure to C that most students get. Not to sound pompous or anything but I knew C before I knew Java, and watching people who are otherwise bright students stumble with it is truly a sight to behold. Lisp is only covered as part of the AI courses (the languages course uses ML to introduce functional programming).
A cirriculum that's not straight out of 1998 would be lovely. This isn't really a problem as far as theory goes, but I'm told this is particularly bad in the software development class (version control is barely covered, for example). And, as I said, a student with no previous experience coming in gets to start with Java - and there's also a Visual Basic class, typically taken by nonmajors, but also those majors who aren't comfortable with the Java class.
Basically we have a really weird balance to strike between practical knowledge and theoretical foundation - and a lot of this stems from the fact that the introductory courses have to cater to CS, engineering, web design, IT, math, and physics majors - I'm not sure what the best way to solve it is. I personally would advocate doing the introduuctory courses in Python and then introducing Java at a later time, but the "everyone uses Java in the real world" mentality is strong.
