Up to Ex34.

ex31/lecture.md

@@ -11,27 +11,54 @@ Common Undefined Behavior
 The Plan
 ====
 
+Review the issues around Undefined and Unspecified Behavior (UB).
+
+
+
+Read The Book
+====
+
+The book lists many of the UB and discusses why they are important to know 
+about.
+
+
 
 The Code
 ====
 
+There is no code for this exercise, just a quick discussion for the book.
 
 
-The Analysis
+
+Undefined Behavior
 ====
 
+* Compiler writers can do whatever they want.
+* This means even *nothing*, which will ruin you silently.
+* It's best to avoid it.
 
 
 
-Breaking It
+Unspecified Behavior
 ====
 
+* For practical purposes unspecified is the same as undefined.
+
+
+
+Handy Tools
+====
+
+* Clang's UB helpful flags.
+* Lint tools and static analyzers.
 
 
 
 Extra Credit
 ====
 
+Spend a day reading through as much of the UB as you can and find examples of each.  Expect lots of frustration and failure when you do this.
+
 
 
 End Of Lecture 31

ex32/lecture.md

@@ -11,10 +11,48 @@ Double Linked Lists
 The Plan
 ====
 
+Learn about your very first data structure:
+
+Double Linked Lists
+----
+
+
+Creating A liblcthw Project
+====
+
+We'll need a project for the rest of the book called *liblcthw*.
+
+
+
+Algorithms and Data Structures
+====
+
+A big step in going from amateur to professional is learning
+about data structures and algorithms.
+
+A double linked list is the easiest one.
+
+
+
+Double Linked Lists Visually
+====
+
+I'll quickly draw some diagrams to show you how they work.
+
+
+
+Double Linked Lists in Python
+====
+
+Now let's see them in Python.
+
+
 
 The Code
 ====
 
+Now we can see the C code.
+
 
 
 The Analysis
@@ -22,16 +60,39 @@ The Analysis
 
 
 
-
-Breaking It
+Improving It
 ====
 
+* You can make ``List_clear_destroy`` more efficient by using
+  ``LIST_FOREACH`` and doing both ``free`` calls inside one
+  loop.
+* You can add asserts for preconditions so that the program isn't given a ``NULL``
+  value for the ``List *list`` parameters.
+
+
+
+Improving It
+====
+
+* You can add invariants that check that the list's contents are always correct,
+  such as ``count`` is never ``< 0``, and if ``count > 0``, then ``first`` isn't NULL.
+* You can add documentation to the header file in the form of comments before
+  each struct, function, and macro that describes what it does.
 
 
 
 Extra Credit
 ====
 
+* Research doubly vs. singly linked lists and when one is preferred over
+  the other.
+* Research the limitations of a doubly linked list.  For example, while they
+  are efficient for inserting and deleting elements, they are very slow for
+  iterating over them all.
+* What operations are missing that you can imagine needing?  Some examples
+  are copying, joining, and splitting.  Implement these operations and write the
+  unit tests for them.
+
 
 
 End Of Lecture 32

ex33/lecture.md

@@ -11,27 +11,80 @@ Linked List Algorithms
 The Plan
 ====
 
+Learn two sorting algorithms for double linked lists.
 
-The Code
+
+
+Bubble Sort
 ====
 
+Explaining how bubble sort works visually.
 
 
-The Analysis
+
+Merge Sort
 ====
 
+Explaining how bubble sort works visually.
+
+
+
+The Unit Test Code
+====
+
+We'll do the unit test for this first since we know what the
+results should be.
+
+
+
+The Implementation
+====
+
+Once we have the test and it's failing we simply implement it.
+
+
+
+Improving It
+====
+
+* The merge sort does a crazy amount of copying and creating lists, so find ways to reduce this.
+* The bubble sort description in Wikipedia mentions a few optimizations. Try to implement them.
+* Can you use the ``List_split`` and ``List_join`` (if you implemented them) to improve merge sort?
+* Go through of all the defensive programming checks and improve the robustness of
+  this implementation, protecting against bad ``NULL`` pointers, and then create
+  an optional debug level invariant that works like ``is_sorted`` does
+  after a sort.
 
 
 
 Breaking It
 ====
 
+* Overload the data structure to hit the worst case time complexity.
+* Give it a bad data structure.
 
 
 
 Extra Credit
 ====
 
+* Create a unit test that compares the performance of the two algorithms.  You'll want to look at ``man 3 time`` for a basic timer function,
+  and run enough iterations to at least have a few seconds of samples.
+* Play with the amount of data in the lists that need to be sorted and see if that changes your timing.
+* Find a way to simulate filling different sized random lists, measuring how long they take. Then, graph the result to see how it compares to the
+  description of the algorithm.
+
+
+
+Extra Credit
+====
+
+* Try to explain why sorting linked lists is a really bad idea.
+* Implement a ``List_insert_sorted`` that will take a given value, and using the ``List_compare``, insert the element at the
+  right position so that the list is always sorted.  How does using this method compare to sorting a list after you've built it?
+* Try implementing the bottom-up merge sort described on the Wikipedia page.  The code there is already C, so it should be easy to
+  recreate, but try to understand how it's working compared to the slower one I have here.
+
 
 
 End Of Lecture 33