When a programming languages has statements, they typically have conventions for:
statement separators;
statement terminators; and
line continuation
A statement separator demarcates the boundary between two separate statements. A statement terminator defines the end of an individual statement. Languages that interpret the end of line to be the end of a statement are called "line-oriented" languages.
"Line continuation" is a convention in line-oriented languages where the newline character could potentially be misinterpreted as a statement terminator. In such languages, it allows a single statement to span more than just one line.
Line continuation is generally done as part of lexical analysis: a newline normally results in a token being added to the token stream, unless line continuation is detected.
Whitespace – Languages that do not need continuations
Ellipsis (as three periods–not one special character)
MATLAB: The ellipsis token need not be the last characters on the line, but any following it will be ignored.[7] (In essence, it begins a comment that extends through (i.e. including) the first subsequent newline character. Contrast this with an inline comment, which extends until the first subsequent newline.)
Fortran 77: A non-comment line is a continuation of the prior non-comment line if any non-space character appears in column 6. Comment lines cannot be continued.
COBOL: String constants may be continued by not ending the original string in a PICTURE clause with ', then inserting a - in column 7 (same position as the * for comment is used.)
TUTOR: Lines starting with a tab (after any indentation required by the context) continue the prior command.
[End and Begin] using normal quotes
C, C++ preprocessor: The string is ended normally and continues by starting with a quote on the next line.
Libraries
This section needs expansion. You can help by adding to it. (December 2009)
To import a library is a way to read external, possibly compiled, routines, programs or packages. Imports can be classified by level (module, package, class, procedure,...) and by syntax (directive name, attributes,...)
The above statements can also be classified by whether they are a syntactic convenience (allowing things to be referred to by a shorter name, but they can still be referred to by some fully qualified name without import), or whether they are actually required to access the code (without which it is impossible to access the code, even with fully qualified names).
Inline comments are generally those that use a newline character to indicate the end of a comment, and an arbitrary delimiter or sequence of tokens to indicate the beginning of a comment.
Block comments are generally those that use a delimiter to indicate the beginning of a comment, and another delimiter to indicate the end of a comment. In this context, whitespace and newline characters are not counted as delimiters. In the examples, the symbol ~ represents the comment; and, the symbols surrounding it are understood by the interpreters/compilers as the delimiters.
Indenting lines in Fortran 66/77 is significant. The actual statement is in columns 7 through 72 of a line. Any non-space character in column 6 indicates that this line is a continuation of the prior line. A 'C' in column 1 indicates that this entire line is a comment. Columns 1 though 5 may contain a number which serves as a label. Columns 73 though 80 are ignored and may be used for comments; in the days of punched cards, these columns often contained a sequence number so that the deck of cards could be sorted into the correct order if someone accidentally dropped the cards. Fortran 90 removed the need for the indentation rule and added inline comments, using the ! character as the comment delimiter.
COBOL
In fixed format code, line indentation is significant. Columns 1–6 and columns from 73 onwards are ignored. If a * or / is in column 7, then that line is a comment. Until COBOL 2002, if a D or d was in column 7, it would define a "debugging line" which would be ignored unless the compiler was instructed to compile it.
Cobra
Cobra supports block comments with "/# ... #/" which is like the "/* ... */" often found in C-based languages, but with two differences. The # character is reused from the single-line comment form "# ...", and the block comments can be nested which is convenient for commenting out large blocks of code.
Curl
Curl supports block comments with user-defined tags as in |foo# ... #foo|.
Lua
Like raw strings, there can be any number of equals signs between the square brackets, provided both the opening and closing tags have a matching number of equals signs; this allows nesting as long as nested block comments/raw strings use a different number of equals signs than their enclosing comment: --[[comment --[=[ nested comment ]=] ]]. Lua discards the first newline (if present) that directly follows the opening tag.
Perl
Block comments in Perl are considered part of the documentation, and are given the name Plain Old Documentation (POD). Technically, Perl does not have a convention for including block comments in source code, but POD is routinely used as a workaround.
PHP supports standard C/C++ style comments, but supports Perl style as well.
Python
The use of the triple-quotes to comment-out lines of source, does not actually form a comment.[21] The enclosed text becomes a string literal, which Python usually ignores (except when it is the first statement in the body of a module, class or function; see docstring).
Elixir
The above trick used in Python also works in Elixir, but the compiler will throw a warning if it spots this. To suppress the warning, one would need to prepend the sigil ~S (which prevents string interpolation) to the triple-quoted string, leading to the final construct ~S""" ... """. In addition, Elixir supports a limited form of block comments as an official language feature, but as in Perl, this construct is entirely intended to write documentation. Unlike in Perl, it cannot be used as a workaround, being limited to certain parts of the code and throwing errors or even suppressing functions if used elsewhere.[22]
Raku
Raku uses #`(...) to denote block comments.[23] Raku actually allows the use of any "right" and "left" paired brackets after #` (i.e. #`(...), #`[...], #`{...}, #`<...>, and even the more complicated #`{{...}} are all valid block comments). Brackets are also allowed to be nested inside comments (i.e. #`{ a { b } c } goes to the last closing brace).
Ruby
Block comment in Ruby opens at =begin line and closes at =end line.
S-Lang
The region of lines enclosed by the #<tag> and #</tag> delimiters are ignored by the interpreter. The tag name can be any sequence of alphanumeric characters that may be used to indicate how the enclosed block is to be deciphered. For example, #<latex> could indicate the start of a block of LaTeX formatted documentation.
Scheme and Racket
The next complete syntactic component (s-expression) can be commented out with #; .
ABAP
ABAP supports two different kinds of comments. If the first character of a line, including indentation, is an asterisk (*) the whole line is considered as a comment, while a single double quote (") begins an in-line comment which acts until the end of the line. ABAP comments are not possible between the statements EXEC SQL and ENDEXEC because Native SQL has other usages for these characters. In the most SQL dialects the double dash (--) can be used instead.
There is a wide variety of syntax styles for declaring comments in source code.
BlockComment in italics is used here to indicate block comment style.
InlineComment in italics is used here to indicate inline comment style.
^Three different kinds of clauses, each separates phrases and the units differently:
serial-clause using go-on-token (viz. semicolon): begin a; b; c end – units are executed in order.
collateral-clause using and-also-token (viz. ","): begin a, b, c end – order of execution is to be optimised by the compiler.
parallel-clause using and-also-token (viz. ","): par begin a, b, c end – units must be run in parallel threads.
^semicolon – result of receding statement hidden, comma – result displayed
^ abFrom the R Language Definition, section 3.2 Control structures: "A semicolon always indicates the end of a statement while a new line may indicate the end of a statement. If the current statement is not syntactically complete new lines are simply ignored by the evaluator."
^Visual Basic .NET does not support traditional multi-line comments, but they can be emulated through compiler directives.
^ abWhile C# supports traditional block comments /* ... */, compiler directives can be used to mimic them just as in VB.NET.
^ abThe line continuation character _ can be used to extend a single-line comment to the next line without needing to type ' or REM again. This can be done up to 24 times in a row.
^Fortran does not support traditional block comments, but some compilers support preprocessor directives in the style of C/C++, allowing a programmer to emulate multi-line comments.[24]