Re: java/2412: jv-scan reports parse error

Re: java/2412: jv-scan reports parse error

[Alexandre Petit-Bianco]

The following reply was made to PR java/2412; it has been noted by GNATS.

From: Alexandre Petit-Bianco <apbianco@cygnus.com>
To: Mika.Riekkinen@tellabs.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/2412: jv-scan reports parse error
Date: Fri, 30 Mar 2001 14:46:06 -0800 (PST)

 Mika.Riekkinen@tellabs.com writes:
 
 > jv-scan reports parse error on the following code.
 
 Thanks. Here's a patch that fixes the problem. I also wrote a patch to
 take care of the other jv-scan related parse errors reported by `make
 check.'
 
 I intend to check this in pretty soon, in the branch and in the trunk.
 
 ./A
 
 2001-03-30  Alexandre Petit-Bianco  <apbianco@redhat.com>
 
 	* parse-scan.y (array_type:): Rewritten.
 	(type_declaration:): `empty_statement' replaces `SC_TK.'
 	(class_member_declaration:): `empty statement' added.
 	(method_body:): Simplified.
 	(static_initializer:): Likewise.
 	(primary_no_new_array:): Use `type_literals.'
 	(type_literals:): New rule.
 	(dims:): Set and update `bracket_count.'
 	Fixes PR java/2412.
 
 Index: parse-scan.y
 ===================================================================
 RCS file: /cvs/gcc/gcc/gcc/java/parse-scan.y,v
 retrieving revision 1.20
 diff -u -p -r1.20 parse-scan.y
 --- parse-scan.y	2000/11/20 22:56:14	1.20
 +++ parse-scan.y	2001/03/30 21:56:13
 @@ -242,14 +243,15 @@ interface_type:
  ;
  
  array_type:
 -	primitive_type OSB_TK CSB_TK
 -|	name OSB_TK CSB_TK
 +	primitive_type dims
  		{
 -		  $$ = concat ("[", $1, NULL);
 +	          while (bracket_count-- > 0) 
 +		    $$ = concat ("[", $1, NULL);
  		}
 -|	array_type OSB_TK CSB_TK
 -		{	
 -		  $$ = concat ("[", $1, NULL);
 +|	name dims
 +		{
 +	          while (bracket_count-- > 0) 
 +		    $$ = concat ("[", $1, NULL);
  		}
  ;
  
 @@ -316,7 +318,7 @@ type_import_on_demand_declaration:
  type_declaration:
  	class_declaration
  |	interface_declaration
 -|	SC_TK
 +|	empty_statement
  ;
  
  /* 19.7 Shortened from the original:
 @@ -393,6 +395,7 @@ class_member_declaration:
  |	method_declaration
  |	class_declaration	/* Added, JDK1.1 inner classes */
  |	interface_declaration	/* Added, JDK1.1 inner classes */
 +|	empty_statement
  ;
  
  /* 19.8.2 Productions from 8.3: Field Declarations  */
 @@ -513,14 +516,12 @@ class_type_list:
  
  method_body:
  	block
 -|	block SC_TK
  |	SC_TK
  ;
  
  /* 19.8.4 Productions from 8.5: Static Initializers  */
  static_initializer:
  	static block
 -|	static block SC_TK	/* Shouldn't be here. FIXME */
  ;
  
  static:				/* Test lval.sub_token here */
 @@ -892,14 +893,7 @@ primary_no_new_array:
  |	field_access
  |	method_invocation
  |	array_access
 -	/* type DOT_TK CLASS_TK doens't work. So we split the rule
 -	   'type' into its components. Missing is something for array,
 -	   which will complete the reference_type part. FIXME */
 -|	name DOT_TK CLASS_TK	       /* Added, JDK1.1 class literals */
 -		{ USE_ABSORBER; }
 -|	primitive_type DOT_TK CLASS_TK /* Added, JDK1.1 class literals */
 -		{ USE_ABSORBER; }
 -|	VOID_TK DOT_TK CLASS_TK	       /* Added, JDK1.1 class literals */
 +|	type_literals
          /* Added, JDK1.1 inner classes. Documentation is wrong
             refering to a 'ClassName' (class_name) rule that doesn't
             exist. Used name instead.  */
 @@ -907,6 +901,17 @@ primary_no_new_array:
  		{ USE_ABSORBER; }
  ;
  
 +type_literals:
 +	name DOT_TK CLASS_TK
 +		{ USE_ABSORBER; }
 +|	array_type DOT_TK CLASS_TK
 +		{ USE_ABSORBER; }
 +|	primitive_type DOT_TK CLASS_TK
 +		{ USE_ABSORBER; }
 +|	VOID_TK DOT_TK CLASS_TK
 +		{ USE_ABSORBER; }
 +;
 +
  class_instance_creation_expression:
  	NEW_TK class_type OP_TK argument_list CP_TK
  |	NEW_TK class_type OP_TK CP_TK
 @@ -960,7 +965,9 @@ dim_expr:
  
  dims:				
  	OSB_TK CSB_TK
 +		{ bracket_count = 1; }
  |	dims OSB_TK CSB_TK
 +		{ bracket_count++; }
  ;
  
  field_access:

Re: java/2412: jv-scan reports parse error

[Alexandre Petit-Bianco]

The following reply was made to PR java/2412; it has been noted by GNATS.

From: Alexandre Petit-Bianco <apbianco@cygnus.com>
To: tromey@redhat.com, Mika.Riekkinen@tellabs.com, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: java/2412: jv-scan reports parse error
Date: Wed, 4 Apr 2001 10:36:04 -0700 (PDT)

 Alexandre Petit-Bianco writes:
 
 > I searched GNATS this afternoon but it was still broken. I'll keep an
 > eye on it.
 
 Well, it turns out that searching for strings in a PR doesn't work at
 the moment, but searching using the state of the PRs as a key still
 works.
 
 The other jv-scan related PR we have is java/1074. I'm going to commit
 the patch (trunk and branch) and close both PRs shortly.
 
 ./A

Re: java/2412: jv-scan reports parse error

[Alexandre Petit-Bianco]

The following reply was made to PR java/2412; it has been noted by GNATS.

From: Alexandre Petit-Bianco <apbianco@cygnus.com>
To: tromey@redhat.com
Cc: Mika.Riekkinen@tellabs.com, gcc-gnats@gcc.gnu.org
Subject: Re: java/2412: jv-scan reports parse error
Date: Sat, 31 Mar 2001 01:57:46 -0800 (PST)

 Tom Tromey writes:
 
 > FYI, Alex: I believe this PR is a duplicate of another PR.  If/when
 > you check in the fix, don't forget to close the other one.
 
 I searched GNATS this afternoon but it was still broken. I'll keep an
 eye on it.
 
 > I wish we had an automatic way of making sure the two parsers stayed
 > in sync.
 
 They do very different things and sometimes, it shows in their
 grammar. But we could definitively do a better job than what we're
 currently doing.
 
 ./A

Re: java/2412: jv-scan reports parse error

[Tom Tromey]

The following reply was made to PR java/2412; it has been noted by GNATS.

From: Tom Tromey <tromey@redhat.com>
To: Mika.Riekkinen@tellabs.com
Cc: gcc-gnats@gcc.gnu.org, Alexandre Petit-Bianco <apbianco@cygnus.com>
Subject: Re: java/2412: jv-scan reports parse error
Date: 30 Mar 2001 17:14:25 -0700

 FYI, Alex: I believe this PR is a duplicate of another PR.
 If/when you check in the fix, don't forget to close the other one.
 
 I wish we had an automatic way of making sure the two parsers stayed
 in sync.
 
 Tom

java/2412: jv-scan reports parse error

[Mika.Riekkinen]