Add a few more teaching points on error strategies

Author: graeme-a-stewart

.gitignore

@@ -5,3 +5,4 @@
 .DS_Store
 .ipynb_checkpoints
 tmp-file.txt
+__pycache__

bootstrap.txt

@@ -0,0 +1,23 @@
+Bootstrap
+=========
+
+If you want to follow along during this tutorial then
+I recommend the Anaconda Python3 distribution:
+
+https://www.anaconda.com/download/
+
+For a quick start on CERN resources try:
+
+ssh [email protected]
+$ source /cvmfs/sft.cern.ch/lcg/views/LCG_93python3/x86_64-centos7-gcc7-opt/setup.sh
+$ ipython
+Python 3.6.3 (default, Dec 18 2017, 11:58:54) 
+Type "copyright", "credits" or "license" for more information.
+
+IPython 5.4.1 -- An enhanced Interactive Python.
+?         -> Introduction and overview of IPython's features.
+%quickref -> Quick reference.
+help      -> Python's own help system.
+object?   -> Details about 'object', use 'object??' for extra details.
+
+In [1]: 

presentation-insights.ipynb

@@ -3344,7 +3344,11 @@
   {
    "cell_type": "code",
    "execution_count": 161,
-   "metadata": {},
+   "metadata": {
+    "slideshow": {
+     "slide_type": "-"
+    }
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -3362,7 +3366,11 @@
   {
    "cell_type": "code",
    "execution_count": 162,
-   "metadata": {},
+   "metadata": {
+    "slideshow": {
+     "slide_type": "fragment"
+    }
+   },
    "outputs": [
     {
      "name": "stdout",
@@ -3401,7 +3409,7 @@
    "source": [
     "This is pretty easy - any python file found in the current directory can be imported as a module, then it becomes available, using the filename as the namespace entry\n",
     "\n",
-    "Actually, the files don't need to live in the currect directory, `$PYTHONPATH` gets searched , or `sys.path` "
+    "Actually, the files don't need to live in the currect directory, `$PYTHONPATH` gets searched (from the shell), or `sys.path` inside Python itself"
    ]
   },
   {
@@ -3496,6 +3504,7 @@
     "* Class methods are defined very like functions, using `def`\n",
     "  * The first parameter is the class instance itself, by convention always called `self`\n",
     "* The special method `__init__` is called when an instance of the class is created (a.k.a. a constructor)\n",
+    "  * (BTW, there are lots of these special `__FOO__` attributes in Python, e.g., `__del__` is your destructor)\n",
     "* All data members of the class are referenced via the object instance, `self`\n",
     "  * `self.counter` is a *data member* of the class\n",
     "  * `counter` would be a plain local variable (watch out!)"
@@ -3778,7 +3787,7 @@
     "\n",
     "On point that might be coming clear to you now is that Python is quite happy to pass *any* objects into function calls of methods\n",
     "\n",
-    "If the passed object has the right properties to work with the call, it works; if not, then something will fail\n",
+    "If the passed object has the right properties to work with the call, it works; if not, then something will fail (this is known in the trade as *Duck Typing*)\n",
     "\n",
     "Two useful functions can be used to inspect a class's providence\n",
     "* `isinstance(obj, classinfo)` returns `True` if the object is an instance of, or derived from, the classinfo class\n",
@@ -4336,6 +4345,52 @@
     "    pass # pass is a very handy bit of python syntax used for supporting an empty code block"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "slideshow": {
+     "slide_type": "subslide"
+    }
+   },
+   "source": [
+    "## LBYL and EAFP\n",
+    "\n",
+    "Programming life could be divided into two strategies for dealing with errors\n",
+    "\n",
+    "* *Look Before You Leap* - check that things are going to be ok first\n",
+    "* *Easier to Ask Forgiveness than Permission* - go for it and clean up if you need to"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "slideshow": {
+     "slide_type": "fragment"
+    }
+   },
+   "source": [
+    "```py\n",
+    "try:\n",
+    "    x = my_dict[\"key\"]\n",
+    "except KeyError:\n",
+    "    # handle missing key\n",
+    "\n",
+    "if \"key\" in my_dict:\n",
+    "    x = my_dict[\"key\"]\n",
+    "else:\n",
+    "    # handle missing key\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "In general Python prefers EAFP - there are a few advantaged (like avoiding some race conditions) and generally the code looks rather cleaner\n",
+    "\n",
+    "However, don't get so carried away that you start to use exceptions as control flow (really, keep them for *exceptional* situations)"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {