So yes if and unless are effectively expressions in the form of statements

They are not.

$ perl -c -e'my $x = if (f()) { g(); } else { h(); }'
syntax error at -e line 1, near "= if"
-e had compilation errors.

$ perl -e'my $x = unless (f()) { g(); } else { h(); }'
syntax error at -e line 1, near "= unless"
-e had compilation errors.
[download]

if statements and unless statements are statements, not expressions.

if (w/o else) and (the always short-circuiting) and are supposed to do the same

They're not, and they don't.

$ perl -c -e'my $x = ( f() and g() );'
-e syntax OK

$ perl -c -e'my $x = ( if (f()) { g() } );'
syntax error at -e line 1, near "( if"
-e had compilation errors.
[download]

-MO=Concise shows opcodes, not operators. An if statement and an and expression often (but not always) use the and opcode, but that doesn't tell you anything relevant to this conversation.

Would you say an if statement in C and a while statement in C are the same thing because they both use the same branch opcode?

In short, you are begging the question. You are presuming an if statement does the same as an and expression to conclude that an if<c> statement does the same as an <c>and expression.

So it looks like they DO the same, except that one of them can't return a value unless enveloped by 'do{}', 'map{}...' etc.

The Perldocs are a bit fuzzy about the definitions of
• statement
• expression
• term
My normal understanding is that a program is a list of STATEMENTS.

They start after a semicolon and end with a semicolon, unless:

• neighboring curlies
• file borders
• anonymous [array]/{hash} brackets °
• ... (Something I forgot, because only Perl can parse Perl)
They mostly correspond to "lines" in speech, but line numbers are only recorded per statement.²

A TERM is a sub-expression inside a statement. That's roughly everything you can put into round parens and terms can be nested. This corresponds to a branch or leaf in the op tree.

In order to use a statement as a term, Perl offers the do {BLOCK} syntax. Other possibilities are sub and eval .

Since statements are otherwise bound to be executed in void context (after a semicolon), we use them for their side effects and don't expect them to return (meaningful) values.

But some do, like if .

You've even cited a passage from one of Larry's³ standard books showing that it's not an accident.

EXPRESSION is a general concept,

• perlsyn explicitly calls statements "expressions"
• perlop lists "expressions in parenthesis" as terms and refers to do "being parsed as term".
So I hope it's clearer now that the difference is mainly in how something is parsed into the syntax tree.

I think part of the confusion comes from Larry using C syntax mixed with Lisp semantics.

Disclaimer: I didn't have time to check all canonical Perldocs and sometimes it's not clear if an english word was used in its general meaning or to denote a Perl concept. ( like "Expressions are expressions", "and and or are operators", and so on...)

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery FootballPerl is like chess, only without the dice}

footnote

°) I probably misread perlop here, my tests don't allow to write a statement into [] or {}

²) multiple statements per line or multi-line statements break the mental analogy

³) ERRATA: rsFalse informed me that "Intermediate Perl was written by Schwartz and others, not including Larry."

What an operator returns is a key aspect of what they do.

But the point was that you can't argue that they must return the same thing because they *otherwise* have a similar effect.