Feature #6074: Allow alias arguments to have a comma

• Status: Rejected
• Assignee: matz (Yukihiro Matsumoto)
• Target version: 2.6

[trans (Thomas Sawyer)]

This is one my biggest pet peeves with Ruby. I am always typing:

alias :foo, :bar

And getting a damn syntax error.

Btw, is there a reason why alias is a keyword and not a method?

[marcandre (Marc-Andre Lafortune)]

• Priority changed from Normal to 3

If that this is your biggest pet peeve, you must really love Ruby!

Just use alias_method instead. The alias keyword is more general and allows other aliases like global variables; also you don't need to give symbols to alias, you can write alias foo bar.

[trans (Thomas Sawyer)]

Yes, I have almost exclusively used #alias_method in the past, but it's always seems rather silly to have to use the longer term. And I bet it would be an easy adjustment to support a comma.

Btw, I really dislike using the non-symbol form of alias as it sticks out under syntax highlighters like a sore thumb.

[mame (Yusuke Endoh)]

Hello,

2012/2/24 Thomas Sawyer [e-mail]:

Btw, is there a reason why alias is a keyword and not a method?

Just historical reason, I guess. A keyword-style alias was
first introduced, and then alias_method was later introduced
as a reflection API for the feature.

I understand the feeling. No keyword was needed for such a
feature. If matz designed it now, he would implement it as
a method, I believe.

Btw, I really dislike using the non-symbol form of alias as it sticks out under syntax highlighters like a sore thumb.

In this regard, you should blame your syntax highlighter?

--
Yusuke Endoh [e-mail]

[ko1 (Koichi Sasada)]

• Category set to core
• Target version set to 2.0.0

I think there are no reason why comma should be rejected.

[nobu (Nobuyoshi Nakada)]

Hi,

(12/02/25 13:51), Koichi Sasada wrote:

I think there are no reason why comma should be rejected.

It's simple.

diff --git a/parse.y b/parse.y
index e47dac4..f55e754 100644
--- a/parse.y
+++ b/parse.y
@@ -979,40 +979,40 @@ stmt_or_begin	: stmt
 		    %*/
 		    }
 
-stmt		: keyword_alias fitem {lex_state = EXPR_FNAME;} fitem
+stmt		: keyword_alias fitem opt_comma {lex_state = EXPR_FNAME;} fitem
 		    {
 		    /*%%%*/
-			$$ = NEW_ALIAS($2, $4);
+			$$ = NEW_ALIAS($2, $5);
 		    /*%
-			$$ = dispatch2(alias, $2, $4);
+			$$ = dispatch2(alias, $2, $5);
 		    %*/
 		    }
-		| keyword_alias tGVAR tGVAR
+		| keyword_alias tGVAR opt_comma tGVAR
 		    {
 		    /*%%%*/
-			$$ = NEW_VALIAS($2, $3);
+			$$ = NEW_VALIAS($2, $4);
 		    /*%
-			$$ = dispatch2(var_alias, $2, $3);
+			$$ = dispatch2(var_alias, $2, $4);
 		    %*/
 		    }
-		| keyword_alias tGVAR tBACK_REF
+		| keyword_alias tGVAR opt_comma tBACK_REF
 		    {
 		    /*%%%*/
 			char buf[2];
 			buf[0] = '$';
-			buf[1] = (char)$3->nd_nth;
+			buf[1] = (char)$4->nd_nth;
 			$$ = NEW_VALIAS($2, rb_intern2(buf, 2));
 		    /*%
-			$$ = dispatch2(var_alias, $2, $3);
+			$$ = dispatch2(var_alias, $2, $4);
 		    %*/
 		    }
-		| keyword_alias tGVAR tNTH_REF
+		| keyword_alias tGVAR opt_comma tNTH_REF
 		    {
 		    /*%%%*/
 			yyerror("can't make alias for the number variables");
 			$$ = NEW_BEGIN(0);
 		    /*%
-			$$ = dispatch2(var_alias, $2, $3);
+			$$ = dispatch2(var_alias, $2, $4);
 			$$ = dispatch1(alias_error, $$);
 		    %*/
 		    }
@@ -5432,6 +5432,10 @@ opt_nl		: /* none */
 		| '\n'
 		;
 
+opt_comma	: /* none */
+		| ','
+		;
+
 rparen		: opt_nl ')'
 		;
 

---

Nobu Nakada

[marcandre (Marc-Andre Lafortune)]

I might have sounded more negative than I intended, so let me say that I also agree that we should allow a comma.

[shevegen (Robert A. Heiler)]

I myself prefer "alias" because:

• It does not require a ','

• It is shorter to type

• It reads easier.

  def bar
  puts 'Hi from bar.'
  end

  alias foo bar
  alias_method :foo, :bar

The first way is more readable for my poor old eyes.

[mame (Yusuke Endoh)]

• Status changed from Open to Assigned
• Assignee set to nobu (Nobuyoshi Nakada)

Sorry, I didn't know alias syntax accepts symbols as arguments!
Then I agree with this proposal.

Ruby always gives me a fresh surprise for me.

--
Yusuke Endoh [e-mail]

[mame (Yusuke Endoh)]

• Target version changed from 2.0.0 to 2.6

[trans (Thomas Sawyer)]

Is there any reason this can't make it into 2.0?

[alexeymuranov (Alexey Muranov)]

Just another idea in this direction: allow

def :foo, :bar do
  puts bar
end

in addition to

def foo(bar)
  puts bar
end

(the "do" in the first case can be optional).

[drbrain (Eric Hodel)]

On Dec 7, 2012, at 9:10, "alexeymuranov (Alexey Muranov)" [e-mail] wrote:

Issue #6074 has been updated by alexeymuranov (Alexey Muranov).

Just another idea in this direction:

If you have a separate idea open a separate feature request.

Do not hijack existing issues.

Hijacking issues makes them difficult for the committers to understand. Since this feature is delayed until next minor, continued off-topic discussion may lead to an issue being closed because its purpose is too difficult to determine.

[nobu (Nobuyoshi Nakada)]

Salvaged. https://github.com/ruby/ruby/pull/609

[nobu (Nobuyoshi Nakada)]

• Description updated (diff)

[nobu (Nobuyoshi Nakada)]

• Assignee changed from nobu (Nobuyoshi Nakada) to matz (Yukihiro Matsumoto)

Request for syntax change.

[knu (Akinori MUSHA)]

I believe the lack of comma is an important sign that alias is not really a method call in which directive arguments are not evaluated like with normal method calls.
If you are allowed to, and should you put a comma in an alias statement, it would look like a method call and confuse you in that you might think it could be passed any expressions.

Here is an example:

class Foo
  name = :to_str
  method = :to_s
  alias name, method  # Likely a mistake, but won't err because `method` happens to exist.
end

I think we should instead deprecate the alias :sym1 :sym2 form so that it is clear that alias takes bare identifiers, not expressions.

[shugo (Shugo Maeda)]

Akinori MUSHA wrote:

I believe the lack of comma is an important sign that alias is not really a method call in which directive arguments are not evaluated like with normal method calls.

Agreed, and this issue should be judged by Matz, shouldn't it?

[matz (Yukihiro Matsumoto)]

• Status changed from Assigned to Rejected

As Akinori stated, alias is special to be special.

In your case, forget alias and always use alias_method.

Matz.

[trans (Thomas Sawyer)]

Special cases always make me cringe. Why is alias a special case anyway? It seems very odd since alias_method is a method and does the same thing.