Slashdot Mirror
Ask Slashdot: How Do You Make Novice Programmers More Professional?
Slashdot reader peetm describes himself as a software engineer, programmer, lecturer, and old man. But how can he teach the next generation how to code more professionally?
I have to put together a three-hour (maximum) workshop for novice programmers -- people with mostly no formal training and who are probably flying by the seat of their pants (and quite possibly dangerous in doing so). I want to encourage them to think more as a professional developer would. Ideally, I want to give them some sort of practicals to do to articulate and demonstrate this, rather than just "present" stuff on best practices... If you were putting this together, what would you say and include?
This raises the question of not only what you'd teach -- whether it's variable naming, modular programming, test-driven development, or the importance of commenting -- but also how you'd teach it. So leave your best answers in the comments. How do you make novice programmers more professional?
This raises the question of not only what you'd teach -- whether it's variable naming, modular programming, test-driven development, or the importance of commenting -- but also how you'd teach it. So leave your best answers in the comments. How do you make novice programmers more professional?
Let them accumulate experience and wisdom, and when they achieve it then tell them they're too old to work in this field.
Ideally, I want to give them some sort of practicals to do to articulate and demonstrate this, rather than just "present" stuff on best practices...
This raises the question of not only what you'd teach -- whether it's variable naming, modular programming, test-driven development, or the importance of commenting -- but also how you'd teach it.
I think this is already going down the wrong path. Those are just technical skills and practices that will be picked up over time, and some kind of workshop isn't really the place to learn them imo.
The important differences between a new guy and someone with a decent bit of professional experience under their belt isn't so much in technical skills or adherence to best practices, but it's more of a mindset and general direction thing. Once you've seen a few projects from start to completion, you start to recognize certain patterns and points where things can go really well or really bad. Once you've worked on a bunch of different teams, you start to recognize how different people contribute to a team dynamic and the various ways in which a team functions. You start to understand how your job integrates with the rest of our department and the rest of the business, how the whole management structure works, and what really drives most technical decisions (hint: technical merit is often the last thing driving a decision).
The problem is, you can't really teach that. So I guess my answer would really be very generic "how to be a good employee" type stuff: Take ownership of your problems, check your ego, play well with others, etc. Being a more professional programmer has little imo to do with being a better programmer and more to do with being a better professional. You become a more professional programmer by learning how to have a productive meeting with management about your project, not by learning the magic of continuous integration.
If they are new programmers, probably they need more than just programming skill, they need skill acting like a professional. The Clean Coder does a really good job with that.
For programming skill, I'm going to suggest Zero Bugs and Program Faster. That book tries to change the way people think about code.
On the practical side, there's no substitute for looking at good code. Assuming you're a good programmer, this would mean code review is one method. Have him review your code and find mistakes. He'll think he's trying to catch you, but he'll learn a lot doing it. Then you can review his code, too.
Another good mentoring technique is unit tests. They show you the kinds of things the programmer is thinking about when they test. So you can look over the tests in code review and say, "hey, you forgot to test this aspect." Ideally you'll want him/her to be thinking of every possible test case, even if he/she doesn't actually write out the test.
Another thing is to treat the younger person with respect. Sometimes if you say, "Oh you did that wrong" they will automatically assume, "he hates me" and put themselves in an adversarial stance, which is not helpful for anyone. Look for things that they do that you really respect, and point them out.
"First they came for the slanderers and i said nothing."
My company hires many young non-degreed self-taught programmers (because that is all we can find). We give them a reading list, and require them to spend about four hours per week doing professional reading and studying on their own time. The books include "Clean Code", "Programming Pearls", "The Pragmatic Programmer", several books on algorithms, code complexity, and books on software engineering such as "The Mythical Man-Month" and "Joel on Software". A lot of it is stuff that they would have learned if they had a CS degree. They can substitute books of their own choosing with pre-approval.
Many of these younglings have matured into great programmers. I hired one guy while he was a junior in high school. He worked for several years, and then decided to go to college, and ended up getting a PhD from Stanford.
because that is all we can find
Is there an implicit "for cheap" at the end there? Because lots of old guys are frequently bellyaching here about how after age 40/50 they can't get any work (and one presumes they know the ropes by then).
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
I agree on the examples and discussion around them.
... a working program (without bugs :) was the goal -- however they got there (rule 1). Negative criticism wasn't allowed -- only constructive criticism (rule 2) -- and thankfully there were no issues with this. My usual teaching phrase was to say "a better way to have done the same thing was ___ because ___", and to remind them that in the end we paid them to write working code.
I taught a class for some new programmers fairly recently out of school. They knew PHP and a few other languages, so they could do basic coding. However, we needed them to know Perl and our app.
I taught 5 classes of 1 hour each, 2 every week. There was about 20 minutes of lecture up front to explain ideas beyond the basics, and they were encouraged to ask questions during that. I also gave them a problem to solve at end of the lessons to complete for next time. The homework at the beginning was solvable in 10-20 lines; the ones at the end took nearly 50 lines. The last ~40 minutes of a session was spent going over the homework and discussing it. Each person presented their answer for everyone to look at and were asked what they found hard to do and why and we talked about solutions with emphasis on the why.
Part of the reason it worked well was because I stressed there was not 1 definitive answer, not even mine
The discussion of the homework allowed them to see how others did it, I also showed my answer though I always went last. We also talked about ways to do it better with a large emphasis on why something was better, trade-offs, etc. I also did my best to point them in the direction of good habits: comments, testing, modularization, maintainability, etc. I also mentioned useful books, most of which have been mentioned by others.
If I ever had to teach something like this again, I believe I'd do it the same way. All of the class members gave me good feedback and said they liked the format.
So, question........how do you ensure that they actually read them?
I ask them what they thought of the book, what they learned from it, and what questions they have that the book didn't answer.
These are self-taught people that passed a rigorous interview process consisting mostly of coding. They want to learn. I have never caught any of them faking their professional development.
Could we get a full list of those books?
This is the list I currently use. I welcome additional recommendations. What CS books have you read recently that you really wished you had read 10 years ago?
Programming:
Clean Code
Code Complete
Programming Pearls
The Pragmatic Programmer
Regular Expressions
Algorithms by Robert Sedgewick
Introduction to Algorithms by Tom Cormen
Hacker's Delight by Henry Warren
Interface design:
Don't Make Me Think
The Design of Everyday Things
Microinteractions
Software Engineering:
The Mythical Man-Month
Joel on Software
Test Driven Development
Theory:
The Turing Omnibus
Deep Learning, by Goodfellow, Bengio, Courville
Concrete Mathematics by Donald Knuth
Physics for Game Developers
Computability, Complexity, and Languages
As you said, the common way of getting software requirements doesn't work too well, and certainly doesn't work *reliably*.
I have a book from the 1970s that describes many of the programs I use every week. They still serve the requirements 40 years later. I'll come back to that set of programs, and how they predicted requirements 40 years down the road, at the end of my post.
Before getting to the 40 year old programs that are still used daily around the world, this topic reminds me of one of the best software design tips that I've been taught. In retrospect it seems obvious, but many programmers haven't thought to do it, and most don't insist on doing it.
90% of the time, you're writing software to better do something that's currently being done some other way. Perhaps you're replacing legacy software, perhaps it's currently being done "manually", people entering data one item at a time. Perhaps you're replacing a paper-based system. Most of the time, you're replacing *some* method of doing the same task.
It logically follows, then, that to fully understand the process, it's requirements and idiosyncrasies, you can watch the people actually doing it. Even better, have them show you how they do it, then try to do the job yourself while they watch and correct you or point out things to be careful of. Take notes during this. Most likely, the way they are using the old system is NOT how it was designed to be used, because the designers of the old system weren't clear on the requirements. But users find a way of meeting their requirements. Watching how they do that shows you what they actually need to get their job done.
Already just by watching them do the task you'll understand the requirements far better than you would by having a meeting with their boss's boss (the common, bad, way to get requirements). After watching them do the task, next ask them two questions:
What about the current process or tools is frustrating for you, or slows you down?
Pretending *anything* is possible, what would your impossible wishes be for this?
The second question often elicits ideas that allow the programmer to say "I can do that, that's easy". Then you begin to glow with heavenly lights because they thought their wish couldn't possibly be granted. Truly, I've done EASY programming tasks that have garnered me a reputation for being able to do the impossible, simply by asking the users what impossible features they wish I could provide. Their conception of what's easy and what's impossible is totally unrelated to what a good programmer can actually do. (You've probably noticed users often think it should be easy for us to do something that's actually nearly impossible. The flip side of that same ignorance is that they think we can't do stuff that we can actually do pretty easily.)
I didn't come up with any of this myself, these aren't my genius ideas and I wouldn't expect anyone else to think of these things. These are things I've been taught along the way, and I wouldn't expect another programmer to think of them, until they are also taught these ideas.
One more thought, or set of thoughts about foreseeing requirements. I was also taught that you can, fairly easily, plan for and program for future requirements without knowing what those future requirements will be. There are two major ways of doing that, both closely related. One is to avoid hardcoding unnecessary limitations. As an example, configuration for my software never has the user provide a configuration value. Instead, each configuration item is a LIST. If my software can send email notifications, it isn't configured with an email address to send to, it's configured with a list of email addreses. If it can read from a data file, it can read from a list of data files, etc. In the code, the added flexibility requires just this additional code:
foreach {
}
That's it. Just "foreach" whenever a configured value is used makes the whole system far more flexible. This is an example of not ar
Project Euler is a test of math skills, not programming skills. The two are often conflated, but in reality overlap only rarely. In many branches of programming you can function very well while knowing nothing more than what +, -, *, and / do.
Moreover, the problems are arranged in chains that lead up from basic understanding to advanced understanding. Simply dumping one at random into the lap of an unprepared person is very likely to weed out all the excellent programmers, leaving only math students - who I'm guessing aren't responding to your ads to begin with. So that's your problem right there: you ask for one skillset, and then you are surprised that your test for another skillset isn't working out. It isn't the quality of your candidates. It's you.
If this wasn't clear enough: I've been programming since the eighties, I have my masters degree, and somehow they trust the software I write controlling (which literally means "deciding life and death of") spacecraft costing 300 million euro and up. If I ever fuck up I guarantee you _will_ read about it here on slashdot. Yet somehow I have _never_ needed to determine if a number is prime, or indeed any of the other circus tricks at Project Euler.
until social scum like you forced the socially awkward out of the industry which they built for you
I'd say that the industry has evolved beyond the basement dwelling coder; social skills are more important in IT these days, and the industry is better for it. But now that there are fewer job openings, employers get to be more picky, and they don't always select on the right criteria. A strong emphasis on social skills where they aren't needed, for example. Or just hiring the most likable of qualifying candidates. Add to that the widespread notion that programming is a young man's game, that good programmers always move on to do other stuff at some point, that you're a loser dinosaur if you're still coding past your 30s. I recently had a chat with some ex-colleagues about hiring practices, and I got the impression that their companies are looking for people who are either college graduates, or experienced programmers who at the very least have developed and launched a wildly successful new OS or social media service. But a middle aged coder who is "merely" very good is suspect: why isn't he an architect or a project manager?
If construction was anything like programming, an incorrectly fitted lock would bring down the entire building...
The problem is that our industry, unlike every other single industry except acting and modeling (and note neither are known for "intelligence") worship at the altar of youth. I don't know the number of people I've encountered who tell me that by being older, my experience is worthless since all the stuff I've learned has become obsolete.
This, despite the fact that the dominant operating systems used in most systems is based on an operating system that is nearly 50 years old, the "new" features being added to many "modern" languages are really concepts from languages that are between 50 and 60 years old or older, and most of the concepts we bandy about as cutting edge were developed from 20 to 50 years ago.
It also doesn't help that the youth whose accomplishments we worship usually get concepts wrong. I don't know the number of times I've seen someone claim code was refactored along some new-fangled "improvement" over an "outdated" design pattern who wrote objects that bare no resemblance to the pattern they claim to be following. (In the case above, the classes they used included "modules" and "models", neither which are part of the VIPER backronym.) And when I indicate that the "massive view controller" problem often represents a misunderstanding as to what constitutes a model and what constitutes a view, I'm told that I have no idea what I'm talking about--despite having more experience than the critic has been alive, and despite graduating from Caltech--meaning I'm probably not a complete idiot.)
Our industry is rife with arrogance, and often the arrogance of the young and inexperienced. Our industry seems to value "cowboys" despite doing everything it can (with the management technique "flavor of the month") to stop "cowboys." Our industry is agist, sexist, one where the blind leads the blind, and seminal works attempting to understand the problem of development go ignored.
How many of you have seen code which seems developed using "design pattern" roulette? Don't know what you're doing? Spin the wheel!
Ours is also one of the fewest industries based on scientific research which blatantly ignores the research, unless it is popularized in shallow books which rarely explore anything in depth. We have a constant churn of technologies which are often pointless, introducing new languages using extreme hype which is often unwarranted as those languages seldom expand beyond a basic domain representing a subset of LISP. I can't think of a single developer I've met professionally who belong to the ACM or to IEEE, and when they run into an interesting problem tend to search Github or Stack Overflow, even when it is a basic algorithm problem. (I've met programmers with years of experience who couldn't write code to maintain a linked list.)
So what do we do?
Beats the hell out of me. You cannot teach if your audience revels in its ignorance and doesn't