All 3 entries tagged Winkt
No other Warwick Blogs use the tag Winkt on entries | View entries tagged Winkt at Technorati | There are no images tagged Winkt on this blog
March 02, 2022
Learning to Program
We (Warwick RSE) love quotes, and we love analogies. We do always caution against letting your analogies leak, by supposing properties of the model will necessarily apply to reality, but nontheless, a carefully chosen story can illustrate a concept like nothing else. This post discusses some of my favorite analogies for the process of learning to program - which lasts one's entire programming life, by the way. We're never done learning.
Learning to program is a lot like trying to complete a jigsaw puzzle without the picture on the box. At first, you have a blank table and a heap of pieces. Perhaps some of the pieces have snippets you can understand already - writing perhaps, or small pictures. You know what these depict, but not where they might fit overall. You start by gathering pieces that somehow fit together - edges perhaps, or writing, or small amounts of distinctive colours. You build up little blocks (build understanding of isolated concepts) and start to fit them together. As the picture builds up, new pieces fit in far more easily - you use the existing picture to help you understand what they mean. Of course, the programming puzzle is never finished, which is where this analogy starts to leak.
I particularly like this one for two reasons. Firstly, one really does tend to build little isolated mini-pictures, and move them around the table until they start to connect together, both in jigsaws and in programming. Secondly, occasionally, you have these little moments - a piece that seemed to be just some weird lines fits into place and you say "oh is THAT what that is!". One gets the same thing when programming - "oh, is that how that works!" or "is that what that means".
Personally, a lot of these moments then become "Why didn't anybody SAY that!". This was the motivation behind our blog series of "Well I Never Knew That" - all those little things that make so much sense once you know.
The other motivation for the WINKT series is better illustrated using a different analogy - hence why we like to have plenty on hand.
A Walk in the Woods
Another way to look at the learning process, is as a process of exploring some new terrain, building your mental map of it. At first, you mostly stick to obvious paths, slowly venturing further and further. You find landmarks and add them to your map. Some paths turn out to link up to other paths, and you start to assemble a true connected picture of how things fit together. Yet still there is terrain just off the paths you haven't gone into. There could be anything out there, just outside your view. Even as you venture further and deeper into the woods, you must also make sure to look around you, and truly understand the areas you've already walked through, and how they connect together.
My badly done keynote picture is below. The paths are thin lines, the red fuzzy marks show those we have explored and the grey shading shows the areas we've actually been into and now "understand". Eventually we notice that the two paths in the middle are joined - a nice "Aha" moment of clarity ensues. But much of our map remains a mystery.
And that is one of the reasons I really like this analogy - all of the things you still don't know - all that unexplored terrain. This too is very like the learning process - you are able to use things once you have walked their paths, even though there may be details about them you don't understand. On the one hand, this is very powerful. On the other, it can lead to "magic incantations" - things you can repeat without actually understanding. This is inevitable in a task this complex, but it is important to go back and understand eventually. Don't let your map get too thin and don't always stick to the paths, or when you do step off them, you'll be lost.
This was the other main motivation behind our WINKT series - the moments when things you do without thinking suddenly make sense, and a new chunk of terrain gets filled in. Like approaching the same bit of terrain from a new angle, or discovering that two paths join, you gain a new perspective, and ideally, this new perspective is simpler than before. Rather than two paths, you see there is only one. Rather than two mysterious houses deep in the woods, you see there is one, seen from two angles.
Take Away Point
If you take one thing away from this post, make it this: the more you work on fitting the puzzle pieces together, the clearer things become. Rather than brown and green and blue and a pile of mysterious shapes, eventually you just have a horse.
And one final thing, if we may stretch the jigsaw analogy a little further: just because a piece seems to fit, doesn't mean it's always the right piece...
October 10, 2019
The Basics aren't so Basic
Sorry for the long break! We've been busy with the start of term and busy expanding our training material (link). This week I am just going to talk about something that you should always keep in mind, not just with programming and computers but with a whole bunch of things, and that is, what does it mean to say something is "basic".
There is a quote often attributed to Einstein, although not directly traceable to himwhich goes
Common sense is the collection of prejudices acquired by age eighteen.
Whether the origin is real or not, it's often true that what people think is simple, or "just common sense" is only so because of their background. To somebody who cut teeth on a BBC Micro, programming might seem super BASIC.
Jokes aside, you will probably keep coming across things that are super-basic and feeling a bit awkward that they somehow escaped you until now. Especially if you have learned things in the usual manner, i.e. by necessity, it is very easy to miss some of the basics. You can find yourself doing really quite advanced things, while not knowing something that "everybody else" seems to. This is normal. It is not beneficial, but it is perfectly normal. Frankly, in computing and programming there is a vast, vast sea of "basics", and no matter how much you learn there always seems to be more.
When I were a Lad
When I was a PhD student, I was happily using 'ssh' to login to remote machines, but I would always type out the whole host spec, such as "firstname.lastname@example.org". I remember feeling a bit dumb when my supervisor pointed out that I didn't need the "username", and he thought this was somehow basic and obvious. I was frankly a little bit irritated because nobody told me! How was I meant to know?
"Simplicity is the final achievement."
(Quote from Frederic Chopin)
Moreover, just because something is "basic" doesn't mean it is simple. In fact, Merriam-Webster's definitionof the adjective "basic", while perhaps a bit unhelpfully recursive does not say simple anywhere. That thing with the username isn't so simple. It's fundamental, sure, but it's not simple!
Years later, I am still regularly coming across things that are "basic" that I have never encountered before. The whole "learning how to program" thing is far more of a helix than a road. You come across fundamental things all the time, some for the first time, some repeatedly, and often you can understand them better every time. Eventually, you find them simple. Sometimes they feel even elegant, because they arise so smoothly from the things you do know, or perhaps even seem so obviously "the only way it could be".
This is most of the motivation for our "WINKT" blog post series. These are fundamental, mostly "basic" things, but they're mostly not things you could usefully be told about the first go-around. Mostly, they are the basics of how the complicated things work. For example:
- On the command line: if you use the '*' wildcard, when does this get expanded into the list of matches? Specifically, if you accidentally create a file called "-rf" in your home, and ran the command `rm *` to remove files, how much trouble would you be in? The answer is, _a lot_. * is processed first, by the shell, and unfortunately '-' comes first in the alphabet. You just ran the equivalent of `rm -rf *`. Ooops.
- Any C/C++ programmers: if you use a variable which is undefined, what is it's value? If you said "whatever is in the associated memory beforehand", you're close, but wrong. An undefined variable is undefined behaviour - it can be given any value, including a different one each time it is accessed. Why? Because the standard says so. But who needs to know that? It is enough to know that its value is unreliable. Using your "basic" knowledge of the C memory model, you would likely guess the above, and it would never matter. [Disclaimer: this is one that I personally only learned a few weeks ago. It's absolutely fundamental, but not at all simple.]
- For Fortran 2003 people: if you have a function-scoped ALLOCATABLE array, allocate it inside the function and forget to free it before the function exits, what happens? A memory leak? Nope! Fortran will helpfully deallocate the array on exit. If you didn't know this and freed everything yourself, there would never be a problem, but this one often surprises people.
- For Python people: suppose you give a function a default argument, like `def func(arg, list_arg = ): ...` and suppose inside the function, list_arg gets filled with stuff. If you call the function twice without supplying list_arg, what do you get the second time? If you said "the combined contents from the first and second calls" you would be correct! The default arg. is an empty list, but it is the SAME empty list each time!
What's the point of all this rambling? Just that there is so much often classed as "the basics" that nobodycan know it all, and there is nothing so basic that you wouldn't do well to re-examine it anyway. It gets said all the time, but with computers there really are no stupid questions. Well, OK, there are some pretty stupid questions. But I have never seen one yet that wasn't worth thinking about.
Postscript: any suggestions for things that make you go "Well I Never Knew That"!, email us or comment! We can always use more
September 10, 2019
New snippets series: WINKT
Super short blog post to start a new snippets series. Along with our SOUP series (Snippet of the <Undefined Period>) we're trying out a new way of posting. WINKT (Well I Never Knew That) is a place for all those things that we stumble across that don't warrant a complete post, but are interesting (if you're a giant geek, anyway).
Fortran Variable Declarations
To start the series off, something I never thought about in Fortran: when you declare a variable with multiple attributes, what is the Left-Hand side comma-separated list actually doing? I was reading a Doctor Fortran post about the potential pitfalls of blindly applying every attribute you might wantand saw his final example of applying (here erroneously) the DIMENSION keyword to a variable in a separate line. "Wait, " I said, "can you do that?" I wrote a quick example, compiled it (with strict standards adherence) and proclaimed "Well I Never Knew That!"
What am I talking about here? Well, consider creating an array of integers. You'll usually write something like:
INTEGER, DIMENSION(:, :), ALLOCATABLE :: I
for a 2-D array. As it turns out, you could equivalently write this as
INTEGER :: I DIMENSION :: I(:, :) ALLOCATABLE :: I
adding the attributes across several lines.
The comma-separated lists of attributes on the left of the double-colon are effectively "stacking" the attributes, but each attribute can also be applied separately.
In that example, it's clearly a bit silly and unwieldy, but it's something you might see "in the wild" and perhaps be confused by, so worth knowing. In some examples, it might actually help make things clear, with, for example, attributes such as SAVE or INTENTs, which can sometimes get "lost in the noise" of declarations. So rather than
INTEGER, SAVE :: I INTEGER :: j, k
I could write
INTEGER :: i, j, k SAVE :: i
This might look clearer, especially if I have more stacked attributes.
This is probably not something I will ever use, and I am not sure I would recommend it, since it looks a bit unexpected, and generally code should avoid unexpectedness. But it did show up to me just how the attribute lists must be working, and was an interesting ten minutes.