On Saturday, 7 June 2014 at 06:48:39 UTC, Philippe Sigaud wrote:
I was watching Chuck Allison talk yesterday, and wondered what
could be a possible homework in D. Maybe other people here have
some ideas, maybe Bearophile will point to RosettaCode, I don't
know. But here is a possible idea:
Trees.
Since you taught them about ranges/lists and functional
mapping/filtering/folding on them, maybe an opening towards
trees could be interesting. Trees are present everywhere as a
structure: documents, initialization files, even source code.
Defining a generic, polymorphic, tree node is easy in D:
struct Tree(T)
{
T value;
Tree[] children;
}
There. Not much more complicated than T[], right?
Have them play with factory functions, assembling trees
together and so on. Writing a way to print a tree on the
console is also fun.
Then, show them how to map a function on a tree, to get a
different tree with the same shape:
Tree!int tree1 = ...
Tree!string tree2 = treeMap!((int i)=> to!string(i))(tree1);
Then, ask them why there is no (easy) way to filter a tree :)
And, reduce. Reducing a tree is very powerful. Getting its
height, the number of nodes. Show to them that the printing
function they wrote to get trees on screen is in fact a reduce.
And now for the fun part, what could be a real assignment (is
that the US word?): encode a document as a tree.
Maybe like this:
enum MarkUp { text, title, section, emphasis, ... }
struct DocNode
{
MarkUp nodeType;
string content;
DocNode[] children;
}
So a doc is a root node, probably a section, containing other
subsection, and so on.
And the assignment is: write different functions to output the
tree content as HTML, XML, DocBook, LaTex, Wikipedia mark-up or
markdown. Of course, that should use map/reduce :) Outputting
something that can be seen on their browser should interest
them.
Of course, if at this stage they already know parsing (from the
compiler course), they can also create the tree by parsing an
input text in a simple markup language, and then converting
into another format, all of that using D. But maybe that's too
complicated?
Thoughts? Do other people here have homework ideas?
I did that for a project at work. A generic tree with specific
DOM implementation. It's a simple concept, yet interesting,
especially when you get a chance to eliminate JS's shortcomings.
I split it up though. I have a generic Tree(T) that takes child
nodes of various types like this:
struct / class Element(T) {
T name;
T[T] attributes;
// ...
string toString() {
return ...;
}
}
So each element manages / can look at its own data, and the
Tree(T) is just an interface that manages these elements and
exposes methods like "getElementsByTagName" etc. I don't know, if
this the best approach but it works fine for me atm.