The following is a simple representation of an xtext grammar with cross references.
There are two entities - a container and an object - and operations on each. I would like to be able to refer to an object either by it's own name, or qualified with the container name. I would like cross-references to be available in either case.
The grammar I have is:
Model:
operations+=Operation*;
ContainerEntity:
name=ID;
ObjectEntity:
(first=[ContainerEntity] '.')? name=ID
;
Operation:
CreateContainer | CreateObject | ContainerOp | ObjectOp
;
CreateContainer:
'Container' container=ContainerEntity ';'
;
CreateObject:
'Object' object=ObjectEntity ';'
;
ContainerOp:
'ContainerOp' name=[ContainerEntity] ';'
;
ObjectOp:
'ObjectOp' name=[ObjectEntity] ';'
;
And the editor statements are:
Container c;
Object o;
ContainerOp c;
ObjectOp o;
ObjectOp c.o; // ERROR: Couldn't resolve reference to ObjectEntity 'c'.
Note that it does not recognize "c" as a ContainerEntity. What can I do to make this work the way I described?
Related
Sybase has that non-ANSI SQL conditional operator used to express outer join: *=.
It's being deprecated (http://dcx.sybase.com/1200/en/dbusage/apxa-transactsqlouter-joins-aspen.html).
As we are moving from Sybase ASE to MySQL I have started to use ANTLR4 to parse the Sybase SQL code to try to translate it into the MySQL code equivalent.
I have tried adding it to the TSqlParser.g4 grammar available here: https://github.com/antlr/grammars-v4/tree/master/sql/tsql. See '*' '=' at the end of the line below but it doesn't work
// https://msdn.microsoft.com/en-us/library/ms188074.aspx
// Spaces are allowed for comparison operators.
comparison_operator
: '=' | '>' | '<' | '<' '=' | '>' '=' | '<' '>' | '!' '=' | '!' '>' | '!' '<' | '*' '='
;
I tried a few things to make it work like escaping \* and removing the *= assignment_operator but nothing works. It's probably a dumb question since I'm new to ANTLR. :-(
Please help.
The input *= is being tokenised as a MULT_ASSIGN by the lexer. You defined it as two separate tokens: '*' '=', which is not the same as '*='.
If you parse following the input with your grammar:
SELECT Q
FROM T
WHERE ID * = 42;
it will go fine, but to parse this properly:
SELECT Q
FROM T
WHERE ID *= 42;
you need to do it like this:
comparison_operator
: ... | '*='
;
and to support both, do this:
comparison_operator
: ... | '*' '=' | '*='
;
I've been developing a DSL as an eclipse plugin using Xtext, and so far I was able to reach up to the point where the runtime eclipse application provides a list of suggestions when pressed Ctrl+space. However, the suggestions are only displayed for the first line of the file. Afterwards, no matter how many times I pressed Ctrl+space, suggestions will not come. Below is my Xtext grammar:
Domainmodel:
(elements+=MainElement)
;
MainElement:
ProjectionName | ProjectionComponent | LayerSpecification |
Description | Capability | Category | ServiceGroup |
IncludeFragment | {MainElement} Override | {MainElement} Overtake
;
ProjectionName:
'projection' modelName=ID ';'
;
ProjectionComponent:
'component' componentName=ID ';'
;
LayerSpecification:
'layer' layerName=ID ';'
;
Description:
'description' string=STRING ';'
;
Capability:
'capability' type=('Online' | 'Offline') ';'
;
Category:
'category' type=('Integration' | 'ExternalB2B' | 'Users') ';'
;
ServiceGroup:
'servicegroup' type=('Mobility' | 'Reporting') ';'
;
IncludeFragment:
('#Dynamic_Component_Dependency' componentName=ID) 'include' 'fragment' fragmentToIncludeName=ID ';'
;
Override:
'#Override'
;
Overtake:
'#Overtake'
;
I have also tried another simpler example (mentioned below) which does not have this problem: -
Domainmodel: (elements+=MainElement)* ;
MainElement: FileName | Type ;
Type: Component | Layer | Description | Category | Entity | Comment ;
FileName: 'projection' name=ID ';' ;
Component: 'component' name=ID ';' ;
Layer: 'layer' name=ID ';' ;
Description: 'description' string=STRING ';' ;
List: Users | Developers ;
Users: 'Users' ;
Developers: 'Developers' ;
Category: 'category' lists=List ';' ;
Entity: 'entityset' name=ID 'for' name2=ID ';' ;
Comment: '----------' comment=ID '----------' ;
Could anyone please help me with understanding why the mentioned problem happens to the first code and not the second?
Thank You!
Your second grammar uses a zero-to-many cardinality in the rule Domainmodel: (elements+=MainElement)* ; whereas your first grammar appears to be missing the * sign Domainmodel: (elements+=MainElement);.
Fixing that one to Domainmodel: (elements+=MainElement)*; will help.
public
class FirstExample { // Here in this example we define some properties in our class
WRITE // This WRITE command follows a rule type defined in my grammar
Hello World
private
String firstTitle ;
private
String secondTitle ;
private
Integer firstAmount ;
private
Integer secondAmount ;
SET
firstTitle = "Ramesh"
SET
secondTitle = "Suresh"
SET
firstAmount = 100
SET
secondAmount = 200
When I change the command 'WRITE' to 'WRITEE' then error highlights in every line.
In eclipse java editor when we write a wrong syntax then error shows on that line only.
I am searching solution for this issue for past 2 days but not able to find the solution.
Kindly help me.I am not able to upload the picture due to some 10 reputation message comes while posting the question..
I am sending the grammar.Kindly check it once.
grammar org.xtext.example.mydsl.MyPoc with org.eclipse.xtext.common.Terminals hidden (GUESS_COMMENT,WS)
generate myPoc "http://www.xtext.org/example/mydsl/MyPoc"
Domainmodel:
(elements+=Type)*;
Type:Class;
// This is the starting point of the grammar
Class:
(packageList+=packageList)*
(packageList+=Import)*
(directives+=Directives) 'class' name=ID ('extends' superType=[Class])? '{'
(greetings+=Greeting)*
(features+=Feature)*
setValues+=SetValues*
operations+=Operation*
functionCall+=ArithmeticOperation*
directives+=Directive*
allMethod+=AllMethod*
allMethodInClass+=AllMethodInClass*
samples+=Samples*
'}';
//Package list consist of all the packages we want to include in our class
packageList:
('package' name=ID) ('.'?) (classes+=[Class])?;
//Import list consist of all the packages we want to include in our class
Import:
'import' importedNamespace=QualifiedNameWithWildcard;
QualifiedNameWithWildcard:
'QualifiedName' '.*'?;
Greeting:
(directives+=Directives)'Hello World'';'
(directives+=Directives)?'bye'?';'
(directives+=Directives)?';'
;
// Features defines the properties of that class
// In our grammar by default we make the property private
// our grammar takes two types of DataType String or Integer
// Name of the property given as per rule ID defined in Terminal file
Feature:
(directives+=Directives)* type=DATA_TYPE name=ID ';';
// In SetValues we initialize the properties by using SET directive
SetValues:
(directives+=Directives) name=ID ('='?) (INT | STRING)';';
//In Operation we define the function,arguments with there Data Type
Operation:
'Do' name=ID ('(' types+=DATA_TYPE args+=ID (',' types+=DATA_TYPE
args+=ID)* ')')?';';
//In Arithmetic Operation we call the function using $ passing the parameters
ArithmeticOperation:
(directives+=Directives)? name=ID '=' (directives+=Directives)? ('$')? types+=[Operation] ('(' (args+=ID) ','
(args+=ID)* ')')?';';
//In Samples we use Do While loop and inside that loop we use if else block
Samples:
name=ID (directives+=Directives) ('!,' STRING)? loopConstruct+=LoopConstruct* operations+=Operation*;
// In AllMethod we can call any method by using class name with '.' operator
AllMethod:
(classes+=[Class]) "." types+=[Operation] ('(' (args+=INT | ID) ',' (args+=INT | ID)* ');')?;
// In AllMethodInClass we can see all methods present in that class
AllMethodInClass:
types+=[Operation] ('(' (args+=INT | ID) ',' (args+=INT | ID)* ');')?;
//In LoopConstruct we define how we can use Do while loop with directive types given below
LoopConstruct:
(directives+=Directives) ("{"
(directives+=Directives ('!,' STRING))*
(directives+=Directive)*
"}")
(directives+=Directives) ("{"
(directives+=Directives ('!,' STRING))*
(directives+=Directive)*
"}")?
(directives+=Directives)?;
Directive:
(directives+=Directives) ('(' name=ID LOOP_CODITION_CHECK name=ID ')')?
'{' (directives+=Directives ('!,' STRING)?)* '}';
Directives:
name=DirectiveType;
DirectiveType:
{DirectiveType} value=DIRECTIVES_TYPE;
// From here all are rules which shows that what we can use in our grammar
terminal DATA_TYPE:
('Integer' | 'String');
// These are some directives
// WRITE is following this rule
terminal DIRECTIVES_TYPE:
('SET' | 'WRITE' | 'READ' | 'QUIT' | 'If' | 'ElseIf' | 'Else' | 'EndIf' | 'Do' | 'while' | 'private' | 'public' | 'do');
terminal LOOP_CODITION_CHECK:
('=' | '>' | '<' | '!')+;
terminal GUESS_COMMENT : '//' !('\n'|'\r')* ('\r'? '\n');
I want to make it clear that WRITE is not only the issue.The line from where grammar breaks it shows error on the remaining lines comes after that line.
After changing my class rule with your code it gives lots of warning.I am sending the warning messages.
warning(200): ../org.xtext.example.myPoc/src-gen/org/xtext/example/mydsl/parser/antlr/internal/InternalMyPoc.g:416:2: Decision can match input such as "RULE_ID {RULE_ID..RULE_DATA_TYPE, RULE_DIRECTIVES_TYPE, 'Do'}" using multiple alternatives: 8, 9
As a result, alternative(s) 9 were disabled for that input
error(201): ../org.xtext.example.myPoc/src-gen/org/xtext/example/mydsl/parser/antlr/internal/InternalMyPoc.g:416:2: The following alternatives can never be matched: 9
warning(200): ../org.xtext.example.myPoc/src-gen/org/xtext/example/mydsl/parser/antlr/internal/InternalMyPoc.g:1166:1: Decision can match input such as "RULE_DIRECTIVES_TYPE '{' RULE_DIRECTIVES_TYPE '!,' RULE_STRING '}' RULE_DIRECTIVES_TYPE" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
warning(200): ../org.xtext.example.myPoc/src-gen/org/xtext/example/mydsl/parser/antlr/internal/InternalMyPoc.g:1184:1: Decision can match input such as "'Do' RULE_ID '(' RULE_DATA_TYPE RULE_ID ',' RULE_DATA_TYPE RULE_ID ')' ';'" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
warning(200): ../org.xtext.example.myPoc/src-gen/org/xtext/example/mydsl/parser/antlr/internal/InternalMyPoc.g:1546:3: Decision can match input such as "RULE_DIRECTIVES_TYPE" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
warning(200): ../org.xtext.example.myPoc.ui/src-gen/org/xtext/example/mydsl/ui/contentassist/antlr/internal/InternalMyPoc.g:568:1: Decision can match input such as "RULE_ID {RULE_ID, RULE_DATA_TYPE..RULE_DIRECTIVES_TYPE, '}', 'Do'..'('}" using multiple alternatives: 8, 9
As a result, alternative(s) 9 were disabled for that input
error(201): ../org.xtext.example.myPoc.ui/src-gen/org/xtext/example/mydsl/ui/contentassist/antlr/internal/InternalMyPoc.g:568:1: The following alternatives can never be matched: 9
warning(200): ../org.xtext.example.myPoc.ui/src-gen/org/xtext/example/mydsl/ui/contentassist/antlr/internal/InternalMyPoc.g:2810:42: Decision can match input such as "RULE_DIRECTIVES_TYPE '{' RULE_DIRECTIVES_TYPE '!,' RULE_STRING '}' RULE_DIRECTIVES_TYPE" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
warning(200): ../org.xtext.example.myPoc.ui/src-gen/org/xtext/example/mydsl/ui/contentassist/antlr/internal/InternalMyPoc.g:2838:39: Decision can match input such as "'Do' RULE_ID '(' RULE_DATA_TYPE RULE_ID ',' RULE_DATA_TYPE RULE_ID ')' ';'" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
warning(200): ../org.xtext.example.myPoc.ui/src-gen/org/xtext/example/mydsl/ui/contentassist/antlr/internal/InternalMyPoc.g:3561:1: Decision can match input such as "RULE_DIRECTIVES_TYPE" using multiple alternatives: 1, 2
As a result, alternative(s) 2 were disabled for that input
After the warning code gives an exception
GeneratorException: (Element: -UNKNOWN-; Reported by: Generator)
org.eclipse.emf.common.util.WrappedException: java.io.FileNotFoundException: ..\org.xtext.example.myPoc\src-gen\org\xtext\example\mydsl\parser\antlr\internal\InternalMyPocLexer.java (The system cannot find the file specified)
I have reduced my grammar and remove all the error and warnings from my code but the error highlighting in every line is still there.
Now my grammar is
grammar org.xtext.example.mydsl.MyPoc with org.eclipse.xtext.common.Terminals hidden (GUESS_COMMENT,WS)
generate myPoc "http://www.xtext.org/example/mydsl/MyPoc"
Domainmodel:
(elements+=Type)*;
Type:Class;
// This is the starting point of the grammar
Class:
(packageList+=packageList)*
(packageList+=Import)*
(directives+=Directives) 'class' name=ID ('extends' superType=[Class])? '{'
(greetings+=Greeting
|features+=Feature
|setValues+=SetValues)*
'}';
//Package list consist of all the packages we want to include in our class
packageList:
('package' name=ID) ('.'?) (classes+=[Class])?;
//Import list consist of all the packages we want to include in our class
Import:
'import' importedNamespace=QualifiedNameWithWildcard;
QualifiedNameWithWildcard:
'QualifiedName' '.*'?;
Greeting:
(directives+=Directives)'Hello World'';'
(directives+=Directives)?'bye'?';'
(directives+=Directives)?';'
;
// Features defines the properties of that class
// In our grammar by default we make the property private
// our grammar takes two types of DataType String or Integer
// Name of the property given as per rule ID defined in Terminal file
Feature:
(directives+=Directives)* type=DATA_TYPE name=ID ';';
// In SetValues we initialize the properties by using SET directive
SetValues:
(directives+=Directives) name=ID ('='?) (INT | STRING)';';
Directives:
name=DirectiveType;
DirectiveType:
{DirectiveType} value=DIRECTIVES_TYPE;
// From here all are rules which shows that what we can use in our grammar
terminal DATA_TYPE:
('Integer' | 'String');
// These are some directives used by us taken from there website
terminal DIRECTIVES_TYPE:
('SET' | 'WRITE' | 'READ' | 'QUIT' | 'If' | 'ElseIf' | 'Else' | 'EndIf' | 'Do' | 'while' | 'private' | 'public' | 'do');
terminal LOOP_CODITION_CHECK:
('=' | '>' | '<' | '!')+;
terminal GUESS_COMMENT : '//' !('\n'|'\r')* ('\r'? '\n');
And in eclipse xtext editor when I write private as privatee it shows error in every line.
My code is
public
class MyPoc {
private
String firstTitle ;
**privatee**
String title ;
private
String secondTitle ;
private
Integer firstAmount ;
private
Integer secondAmount ;
SET
firstTitle = "Ramesh";
SET
secondTitle = "Suresh";
SET
firstAmount = 100;
SET
secondAmount = 200;
}
Now I add some more grammar in my class attribute.The console shows no warning and no exception.But in eclipse editor again when we break grammar syntax it shows error in every line.
Class:
(packageList+=packageList)*
(packageList+=Import)*
(directives+=Directives) 'class' name=ID ('extends' superType=[Class])? LBRACKET
(directives+=Directives* (features+=Feature | setValues+=SetValues | operations+=Operation | functionCall+=ArithmeticOperation | allMethod+=AllMethod
| allMethodInClass+=AllMethodInClass | samples+=Samples) SEMICOLON)*
RBRACKET;
Kindly check it once and correct me if I am doing something wrong.
I have found that problem is not in the class part.The problem is in the parts where we define the rule in the grammar.
Like for ArithmeticOperation
ArithmeticOperation:
(directives+=Directives)? name=ID '=' (directives+=Directives)? ('$')? types+=[Operation] ('(' (args+=ID) ','
(args+=ID)* ')')?';';
Actually I have to write this code in eclipse.Here Concat is the opeartion name means method name.
SET Result = WRITE $ Concat (firstTitle , secondTitle)
Kindly correct me if I ma doing something wrong to achieve my target code.
Regards
Xtext error recovery system is not the same than Java Editor (JDT). Its behavior depends on how you write your rules.
Maybe your grammar is too much ambiguous and you should add a line termination symbol.
Also, you should read this article about error recovery with Xtext: Parser error recovery
Edit
Currently the class rule is very ambiguous with chaining rules with the same first token. The first thing you should do is to rewrite your class rule and remove all ambiguities.(warnings & erros when compiling grammar).
See the following example of class rule:
Class:
(packageList+=packageList)*
(packageList+=Import)*
(directives+=Directives) 'class' name=ID ('extends' superType=[Class])? '{'
(greetings+=Greeting | features+=Feature | setValues+=SetValues
| operations+=Operation | functionCall+=ArithmeticOperation
| directives+=Directive | allMethod+=AllMethod
| allMethodInClass+=AllMethodInClass | samples+=Samples)*
'}';
Edit 2
I made some refactorings to your grammar. And now it has the behavior you asked for.
Domainmodel:
(elements+=Type)*;
Type:Class;
// This is the starting point of the grammar
Class:
(packageList+=packageList)*
(packageList+=Import)*
(directives+=Directives) 'class' name=ID ('extends' superType=[Class])? LBRACKET
(directives+=Directives* (features+=Feature | setValues+=SetValues) SEMICOLON)*
RBRACKET;
terminal SEMICOLON:
';'
;
terminal LBRACKET:
'{'
;
terminal RBRACKET:
'}'
;
//Package list consist of all the packages we want to include in our class
packageList:
('package' name=ID) ('.'?) (classes+=[Class])?;
//Import list consist of all the packages we want to include in our class
Import:
'import' importedNamespace=QualifiedNameWithWildcard;
QualifiedNameWithWildcard:
'QualifiedName' '.*'?;
// Features defines the properties of that class
// In our grammar by default we make the property private
// our grammar takes two types of DataType String or Integer
// Name of the property given as per rule ID defined in Terminal file
Feature:
type=DATA_TYPE name=ID ;
// In SetValues we initialize the properties by using SET directive
SetValues:
name=ID ('='?) (INT | STRING);
Directives:
name=DirectiveType;
DirectiveType:
{DirectiveType} value=DIRECTIVES_TYPE;
// From here all are rules which shows that what we can use in our grammar
terminal DATA_TYPE:
('Integer' | 'String');
// These are some directives used by us taken from there website
terminal DIRECTIVES_TYPE:
('SET' | 'WRITE' | 'READ' | 'QUIT' | 'If' | 'ElseIf' | 'Else' | 'EndIf' | 'Do' | 'while' | 'private' | 'public' | 'do');
terminal LOOP_CODITION_CHECK:
('=' | '>' | '<' | '!')+;
terminal GUESS_COMMENT : '//' !('\n'|'\r')* ('\r'? '\n');
Details of what i changed:
- Removed (useless?) greeting rule.
- Made line/block separators terminals.
- Moved directive to parent rule.
Sample code for editor:
public
class MyPoc {
private
String firstTitle ;
privatee
String title ;
private
String secondTitle ;
private
Integer firstAmount ;
private
Integer secondAmount ;
SET
firstTitle = "Ramesh";
SET
secondTitle = "Suresh";
SET
firstAmount = 100;
SET
secondAmount = 200;
}
I'm trying to define a rule where one element is the perfect candidate for the common terminal rule ID. When I launch the editor to test with sample code, the ID rule is not matched for the string "ABC":
Define : ABC : Holder_1
I get "mismatched input 'ABC' expecting RULE_ID".
Is there something in my grammar conflicting with the ID rule to cause this error?
This is my grammar file:
grammar com.testco.Test with org.eclipse.xtext.common.Terminals
import "http://www.eclipse.org/emf/2002/Ecore" as ecore
generate defwiz "http://www.testco.com/Test"
Define_Holder:
'Definition' ':' holder_name=ID ':' holder_number=HOLDER_NUMBER (':' attribute=define_attr)? (':' pad=holder_pad)?
;
holder_pad:
HEX
;
Type:
TYPE_TERM ':' type_value=TYPE_VAL
;
//***************** TERMINALS *****************
terminal TYPE_TERM:
'Type_' INT+
;
terminal PROFILE:
(PROFILE_ID)(PROFILE_ID)'.'(PROFILE_ID)(PROFILE_ID)
;
terminal HOLDER_NUMBER returns ecore:: EString:
'Holder_' INT+;
terminal HEX returns ecore:: EString :
('0'..'9'|'A'..'F')
;
terminal PROFILE_ID : '^'?('a'..'z'|'A'..'Z') ('a'..'z'|'A'..'Z'|'0'..'9')*;
//***************** ENUMS *****************
enum define_attr:
BINARY='Binary' |
SCRAMBLE='Scramble' |
FORCESIZE='ForceSize' |
FIXEDSIZE='FixedSize'
;
Your rule PROFILE_ID shadows the ID rule for almost every case (except for the underscore).
Yes, it's likely that lexer scans ABC as HEX terminal. Try to define the latter, for example, as follows:
terminal HEX returns ecore:: EString :
'0x' ('0'..'9'|'A'..'F')
;
I have a class Test_Class in Ada 2005 which has a parent-linked task property called Primary, from type Primary_Task, defined as:
type Test_Class is tagged limited
record
Info : Integer;
Value : Float;
Primary : Primary_Task (Test_Class'Access);
end record;
I need build a one-step constructor for my class in the form
function Construct (T : access Test_Class) return Test_Class_Ptr is
begin
return new Test_Class'(Info => T.Info + 1,
Value => 0.0,
Primary => [WHAT I WANNA KNOW]);
end Construct;
Currently my code is:
-- test_pkg.ads
package Test_Pkg is
type Test_Class;
type Test_Class_Ptr is access all Test_Class;
task type Primary_Task (This_Test : access Test_Class) is
pragma Storage_Size (1000);
end Primary_Task;
type Test_Class is tagged limited
record
Info : Integer;
Value : Float;
Primary : Primary_Task (Test_Class'Access);
end record;
function Construct (T : access Test_Class) return Test_Class_Ptr;
end Test_Pkg;
-- test_pkg.adb
with Text_IO; use Text_IO;
package body Test_Pkg is
[...]
function Construct (T : access Test_Class) return Test_Class_Ptr is
T_Ptr : constant Test_Class_Ptr := new Test_Class;
begin
T_Ptr.Info := T.Info + 1;
T_Ptr.Value := 0.0;
return T_Ptr;
end Construct;
end Test_Pkg;
So, how can I code it? What should I put in Primary => [...] code? Should I change the definition of Primary : Primary_Task (Test_Class'Access); in Test_Class definition?
I got an answer from Randy Brukardt (thank you) on comp.lang.ada:
In Ada 2005 or later, use "<>" to default initialize a component in an
aggregate (which is the only thing you can do with a task).
(...)
function Construct (T : access Test_Class) return Test_Class_Ptr is
begin
return new Test_Class'(Info => T.Info + 1,
Value => 0.0,
Primary => <>);
end Construct;
However, I tried to compile it using GNAT GPL 2011 and got the GNATBUG below
c:\tst>gnatmake -gnat12 test_pkg.adb
gcc -c -gnat12 test_pkg.adb
+===========================GNAT BUG DETECTED==============================+
| GPL 2011 (20110428) (i686-pc-mingw32) GCC error: |
| in create_tmp_var, at gimplify.c:505 |
| Error detected around test_pkg.adb:20:29 |
| Please submit a bug report by email to report#adacore.com. |
| GAP members can alternatively use GNAT Tracker: |
| http://www.adacore.com/ section 'send a report'. |
| See gnatinfo.txt for full info on procedure for submitting bugs. |
| Use a subject line meaningful to you and us to track the bug. |
| Include the entire contents of this bug box in the report. |
| Include the exact gcc or gnatmake command that you entered. |
| Also include sources listed below in gnatchop format |
| (concatenated together with no headers between files). |
| Use plain ASCII or MIME attachment. |
+==========================================================================+
Please include these source files with error report
Note that list may not be accurate in some cases,
so please double check that the problem can still
be reproduced with the set of files listed.
Consider also -gnatd.n switch (see debug.adb).
test_pkg.adb
test_pkg.ads
raised TYPES.UNRECOVERABLE_ERROR : comperr.adb:423
gnatmake: "test_pkg.adb" compilation error
So GNAT GPL users may have to wait for the next release to use this solution.