Update CodingStyle.md: placement of *, & and && declarators

Signed-off-by: Berthold Stoeger <bstoeger@mail.tuwien.ac.at>

CodingStyle.md

@@ -20,7 +20,6 @@ other editors that implement this coding style, please add them here.
   continuation lines that are aligned with tabs and then spaces
 
 * all keywords followed by a '(' have a space in between
-
   ```
 	if (condition)
 
@@ -28,7 +27,6 @@ other editors that implement this coding style, please add them here.
   ```
 
 * function calls do NOT have a space between their name and argument
-
   ```
 	i = some_function(argument);
   ```
@@ -41,7 +39,6 @@ other editors that implement this coding style, please add them here.
 
 * all other opening curly braces follow at the end of the line, with a
   space separating them:
-
   ```
 	if (condition) {
 		dosomething();
@@ -50,7 +47,6 @@ other editors that implement this coding style, please add them here.
   ```
 
 * both sides of an if / else clause either use or do not use curly braces:
-
   ```
 	if (condition)
 		i = 4;
@@ -66,13 +62,11 @@ other editors that implement this coding style, please add them here.
   ```
 
 * use space to make visual separation easier
-
   ```
 	a = b + 3 + e / 4;
   ```
 
 * continuation lines have the operator / comma at the end
-
   ```
 	if (very_long_condition_1 ||
 	    condition_2)
@@ -83,7 +77,6 @@ other editors that implement this coding style, please add them here.
 
 * in a C++ constructor initialization list, the colon is on the same line and
   continuation lines are aligned as the rule above:
-
   ```
 	ClassName::ClassName() : x(1),
 		y(2),
@@ -93,14 +86,11 @@ other editors that implement this coding style, please add them here.
   ```
 
 * unfortunate inconsistency
- * C code usually uses underscores to structure names
-
+  - C code usually uses underscores to structure names
   ```
 	variable_in_C
   ```
-
- * C++ code usually uses camelCase
-
+  - In contrast, C++ code usually uses camelCase
   ```
 	variableInCPlusPlus
   ```
@@ -110,7 +100,6 @@ other editors that implement this coding style, please add them here.
 
 * switch statements with blocks are a little bit special (to avoid indenting
   too far)
-
   ```
 	switch (foo) {
 	case FIRST:
@@ -130,7 +119,22 @@ other editors that implement this coding style, please add them here.
   In C code we really like them to be at the beginning of a code block,
   not interspersed in the middle.
   in C++ we are a bit less strict about this - but still, try not to go
-  crazy.
+  crazy. Notably, in C++ the lifetime of a variable often coincides with the
+  lifetime of a resource (e.g. file) and therefore the variable is defined
+  at the place where the resource is needed.
+
+* The `*`, `&` and `&&` declarators are grouped with the name, not the type
+ (classical C-style) as in `char *string` instead of `char* string`. This
+  reflects the precedence rules of the language: `int &i` means that the name
+  `i` stands for a reference [to an object with type `int`], not that
+  `i` stands for an object of the type [reference to `int`].
+  Although this may seem like hairsplitting (both interpretations
+  have the same effect) it is crucial in the
+  definition of multiple variables, such
+  as
+	```
+	struct dive *next, **pprev;
+	```
 
 * In C++ code, we generally use explicit types in variable declarations for clarity.
   Use `auto` sparingly and only in cases where code readability improves.
@@ -157,12 +161,11 @@ other editors that implement this coding style, please add them here.
   spelling and terminology.
   User-visible strings should be passed to the tr() function to enable
   translation into other languages.
-
- * like this
+  - like this
     ```
 	QString msgTitle = tr("Submit user survey.");
     ```
- * rather than
+  - rather than
     ```
 	QString msgTitle = "Submit user survey.";
     ```
@@ -170,7 +173,7 @@ other editors that implement this coding style, please add them here.
   This works by default in classes (indirectly) derived from QObject. Each
   string to be translated is associated with a context, which corresponds
   to the class name. Classes that are not derived from QObject can generate
-  the tr() functions by using the Q_DECLARE_FUNCTIONS macro:
+  the tr() functions by using the `Q_DECLARE_FUNCTIONS` macro:
   ```
 	#include <QCoreApplication>
 
@@ -202,7 +205,7 @@ other editors that implement this coding style, please add them here.
   when the particular translation string is encountered for the first time.
   It remains valid during the whole application's life time.
 
-  Outside of function context, the QT_TRANSLATE_NOOP macro can be used as in
+  Outside of function context, the `QT_TRANSLATE_NOOP` macro can be used as in
   ```
   struct ws_info_t ws_info[100] = {
 	{ QT_TRANSLATE_NOOP("gettextFromC", "integrated"), 0 },
@@ -214,7 +217,7 @@ struct ws_info_t ws_info[100] = {
   ```
   Note that here, the texts will be scheduled for translation with the "gettextFromC"
   context, but the array is only initialized with the original text. The actual
-  translation has to be performed later in code. For C-code, the QT_TRANSLATE_NOOP
+  translation has to be performed later in code. For C-code, the `QT_TRANSLATE_NOOP`
   macro is defined in the "core/gettext.h" header.
 
 * UI text style
@@ -236,11 +239,9 @@ struct ws_info_t ws_info[100] = {
 
 
 * string manipulation
-
  * user interface
     In UI part of the code use of QString methods is preferred, see this pretty
     good guide in [QString documentation][1]
-
  * core components
     In the core part of the code, C-string should be used.
     C-string manipulation is not always straightforward specifically when
@@ -248,7 +249,6 @@ struct ws_info_t ws_info[100] = {
     to help with this. Documentation and usage examples can be found in
     [core/membuffer.h][2]
 
-
 ## Sample Settings
 
 ### Emacs