This Chapter features ECL Error Codes 1 no resulting expression tree 2 "there is an expression tree, but should not be executed" 3 1001 Error in macro expansion (see error follows) This warning indicates there is an error in the macro expansion. Since the compiler will issue an error in the macro, this warning is used only for informational purposes; it gives the position where the macro is called to help the user trace down the problem. Solution: Resolve the error that follows this warning first. Example: 01: f(x) := MACRO 02: x+3; 03: ENDMACRO; 04: 05: EXPORT aaa := f(4); In line 1-3, a value macro f is defined. This is a value macro because it has no Attribute defini- tions. That means the macro body must be legal to appear wherever a value expression may appear. Clearly, the semicolon at the end of line 2 is extraneous, so this warning saying that there is an error in the macro expansion at line 5 appears, and an error following this warning says the error is in line 2 of the macro. 1004 LOCAL specified on dataset that is not Distributed In a SORT or JOIN function, LOCAL can be specified to indicate that the sort or join should be performed while maintaining the distribution of a previously distributed dataset. A dataset is consided to be distributed if: • it has been explicitly distributed as the result of the DISTRIBUTE function • it is defined as already distributed by the DISTRIBUTED function Solution: Use DISTRIBUTE or DISTRIBUTED to distribute the dataset, or remove the LO- CAL option. Example: 01: aaa := DATASET(‘aaa’,{STRING1 f1}, FLAT); 02: distibuted_aaa := DISTRIBUTE(aaa, RANDOM()); 03: distibuted_aaax := DISTRIBUTED(aaa, RANDOM()); 04: 05: OUTPUT(SORT(aaa, f1, LOCAL)); 06: OUTPUT(SORT(distibuted_aaa, f1, LOCAL)); 07: OUTPUT(SORT(distibuted_aaax, f1, LOCAL)); There are 3 datasets defined (in lines 1, 2, 3). We use the LOCAL option to sort them (in lines 5, 6, 7). The first one is not distributed, therefore we get this warning in the SORT on line 5. The second is the result of the DISTRIBUTE function, and the third is specified as already DIS- TRIBUTED. Therefore, lines 6 and 7 compile without warning or error. 1005 Input is explicitly Distribited but Local not specified In a SORT or JOIN function, the dataset(s) involved is(are) distributed, but LOCAL is not explicitly specified. See Warning C1004 for information on distributed. Solution: Add the LOCAL option, or make sure the dataset is not distributed. Example: 01: aaa := DATASET(‘aaa’,{STRING1 fa; }, FLAT); 02: distibuted_aaa := DISTRIBUTE(aaa, RANDOM()); 03: distibuted_aaax := DISTRIBUTED(aaa, RANDOM()); 04: 05: OUTPUT(SORT(distibuted_aaa, fa)); 06: OUTPUT(SORT(distibuted_aaax, fa)); On lines 5 and 6, distributed_aaa and distributed_aaax are distributed, but no LOCAL option is specified in the SORT. Adding ,LOCAL to the SORT in lines 5 and 6 will fix this problem. 1006 CASE does not have any condition No condition is given for the CASE statement. This makes the CASE statement unnecessary under the normal situation. This is allowed for easy use for some special circumstances, like automatic code generation using ECL template language. Solution: Make sure this is what you wanted. Example: INTEGER i := 10; INTEGER j := CASE(i, 3); for other reasons. 1007 Module occurs twice in scope 1010 Unsupported language feature The language feature for which this warning appears is not supported yet. However, the parser can recognize this feature. This means either it is under development, or is deliberately left out Solution: Make sure your QueryBuilder program is the latest version. The feature may be sup- ported already in a newer version than the one your are using. Otherwise, find a workaround solution for you problem. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa; }, FLAT); 02: a1 := aaa(fa = ‘a’); 03: a2 := aaa(fa = ‘b’); 04: OUTPUT(a1-a2); We get this error on line 4 if the dataset difference operation is not supported yet. NOTE: It may be already supported when you read this! 1012 ';' is not legal here (empty statement is not allowed) 1013 A field in transform already assigned a value 1014 physicalLength is not needed, ignored 1015 Module is already imported 1016 Syntax obsoleted: use alternative 1017 Empty assignall has no effort 1018 Unsupported service attribute 1019 An service attribute does not need a value 1020 Module name mismatch 1021 OUTPUT record contains no variables - probably an error 1022 Condition is always true 1023 Condition is always false 1024 #WARNING statement 1025 Duplicate case values 1026 Local will be ignored for a grouped dataset 1027 Local expected 1028 Grouping will be ignored. 1029 No longer used (this is now an error, not a warning) 1030 Counter not used in transform 1031 Merge inputs don't appear to be sorted 1032 Merge inputs don't appear to match 1033 Definition is sandboxed 1034 Definition reference sandbox 1035 regular expression syntax 1036 deprecated syntax 1037 group input does not appear to be sorted 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 2000 2001 Type mismatch - Real value expected In the context where the error occurs, a real value is expected. A real value can be an real con- stant, or an expression that evaluates to a real value. Therefore, 12, 12.34, LOG(0.4)+3, are valid, but ‘12.34’ and ‘String’ are not. Solution: Make sure the expression in question evaluates to a real number. A string constant, like ‘123.2’ is NOT a real number. For C programmers, don’t think ‘C’ or ‘1’ is a char, and therefore can be used as an integer (or real); there is no char type in ECL. Type casting may be needed in some cases. For example: (REAL4)‘12.34’ creates a valid real value by casting the strinf to real. Checking the code before and after the offending code often helps. Example: 01: v1 := EXP(‘STR’); 02: v2 := LN(‘STR’); 03: v3 := LOG(‘STR’); 04: v4 := POWER(‘STR’, 12); 05: v5 := ROUND(‘STR’); 06: v6 := ROUNDUP(‘STR’); 07: v7 := SQRT(‘STR’); 08: v8 := TRUNCATE(‘STR’); 09: v9 := REALFORMAT(‘STR’,3,4); These examples all need to change to a real value. All the functions expect real value as the first parameters. 2002 Type mismatch – Integer value expected An integer value is expected in the context. An integer value can be an integer constant, or an expression that evaluates to an integer value The following is a (not necessarily complete) list where an integer value is expected: • BIG_ENDIAN value • LITTLE_ENDIAN value • WAIT (value) • expr1 op expr2, where op can be: %, DIV, &, |, ^ • substring range: [ start_index .. end_index] • CHOOSE(expr, choose_list) • CHOOSEN(dataset, value) • DEDUP(…, KEEP value) • INTFORMAT(value1, value2, value3) • REALFORMAT(real_value, value1, value2) All bold expressions must be integer values. Solution: Make sure the value is an integer value. Use type cast if necessary. Example: 01: WAIT(‘STR’); 02: v1 := 10.2 % 11.2; 03: v2 := 10.2 DIV 11.2; 04: v3 := 10.2 | 11.2; 05: v4 := 10.2 & 11.2; 06: v5 := 10.2 ^ 11.2; 07: STRING s := ‘This is a string.’; 08: s[0.23]; 09: s[0.23..2.34]; 10: s[..2.3]; 11: s[2.3..]; 12: INTFORMAT(1.2, 2.2, 3.3); 13: REALFORMAT(1.0, 2.0, 3.0); 14: CHOOSE(3.2, 9, 8, 7, 6, 5); 15: aaa := DATASET(‘aaa’,{STRING1 fa; }, FLAT); 16: OUTPUT(choosen(aaa,10.2)); 17: OUTPUT(DEDUP(aaa, fa = ‘A’, KEEP 2.2)); All bold expressions cause such an error. 2003 Type mismatch - String value expected A string value is expected. A string value can be a string constant, or an expression that evaluates to a string value. The following is a list that a string is expected: • EVENT(expr1, expr2) • LENGTH(value) • TRIM(value [, options]) All bold expressions must be string values. For C/C++/Java/Pascal etc programmers, string constants are enclosed in single quotes (‘’) in ECL, not double quotes (“”). Also, ECL currently only supports two escape sequences: \’ and \\ which stands for the single quote and backslash. Support for more escape sequences may be added as needed in the future. Solution: Make sure the expression is a string constant or evaluates to a string value. Use type cast or TRANSFER as needed. Example: 01: WAIT(EVENT(123, TRUE)); 02: len := LENGTH(123); 03: STRING s:= TRIM(123); 2004 Type mismatch - Integer or real value expected An integer or real value is expected. It can be an integer constant, real constant, or an expression that evaluates to a real or integer value. The following is a list of cases where an integer or real value is expected: • expr1 + epxr2 • expr1 – expr2 • POWER(value1, value2) • AVE(dataset, expr) • AVE(GROUP, expr) • DISTRIBUTE(dataset, value) • DISTRIBUTED(dataset, value, tranform) This error is the same as error C2072. Solution: Make sure it is a integer or real value. Use type casting if necessary. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa; }, FLAT); 02: i := 3 + TRUE; 03: j := 4 - TRUE; 04: x1 := POWER(20, TRUE); 05: x2 := AVE(aaa, ‘string’); 06: x3 := AVE(GROUP, ‘string’); 07: d1 := DISTRIBUTE(aaa, ‘abc’); 08: d2 := DISTRIBUTED(aaa, TRUE); All bold expressions are expected to be integer or real value. 2005 Table type mismatch ??? Solution: Example: 2006 Record type mismatch ??? Solution: Example: 2007 Incompatible types There is a type incompatiblity problem here. This normally a result of some other error, like “Integer value is expected”. Solution: Resolve the error before this one (if any) first, otherwise, make sure the types are compatible. Example: 01: i := 3 + TRUE; 02: STRING20 s := ‘ABC’; 03: TYPEOF(s) y := 10; First, an Error C2004 will be reported by the compiler, then this error will be reported. For line 3, since TYPEOF(s) is STRING20, so we such an error for assignment :=10. 2008 Type does not require parameters No parameter is required for a user defined type. A user defined type can be defined in ECL using TYPE. A user defined type can have parameters. See Example below. Solution: Try to get rid of the parameter(s). Example: 01: MyType := TYPE 02: EXPORT INTEGER Load(INTEGER x) := x+1; 03: EXPORT INTEGER Store(INTEGER x) := x-1; 04: END; 05: 06: MyRec := RECORD 07: MyType(3) abc; 08: END; 09: 10: NeedC(INTEGER len) := TYPE 11: EXPORT String Load(String S) := ‘C’+S[1..len]; 12: EXPORT String Store(String s) := S[2..len+1]; 13: EXPORT Integer PhysicalLength(String s) := len; 14: END; 15: 16: Rec := RECORD 17: NeedC(3) good; 18: NeedC(3,5) abc; 19: NeedC xyz; 20: END; In line 7, this error ocurrs. Line 10-14 shows an example that defines a user type with parameter. Line 17 is fine, but an error C2061 occurs in line 18, and C2062 occurs in line 19. 2009 Boolean/Set of/Integer/Real/String/Table/Record/Unknown value expected A certain type is expected to be passed as the parameter in this context. Solution: Make sure the passed parameter is one of the expected types. 2010 Illegal operand for TYPEOF The operand for TYPEOF must be an expression whose type is known. In ECL, we can get the type of an expression and use it as type. All the following are legal uses of TYPEOF: TYPEOF(‘ABC’) x1 := ‘XYZ’; TYPEOF(5) x2 := 10; STRING20 s := ‘ABC’; TYPEOF(s) x3 := ‘DEF’; TYPEOF(x1) x4 := ‘GHI’ Solution: Make sure the type of the operand is known. 2011 Cannot nest TYPE declarations TYPE defintions cannot be nested. Solution: Define these TYPEs separately instead of nested. Example: ReverseString4 := TYPE MyType := TYPE EXPORT INTEGER Load(INTEGER x) := x+1; EXPORT INTEGER Store(INTEGER x) := x-1; END; END; SHARED STRING4 Rev(String4 S) := s[4]+s[3]+s[2]+s[1]; EXPORT String4 Load(String4 s) := Rev(s); EXPORT String4 Store(String4 s) := Rev(s); Here, inside the TYPE defintion for ReverseString4, we are trying to define another TYPE MyType. This nesting is not allowed. 2012 Datasets must be the same type The datasets have different types. The + and – operations can be applied to datasets and these two datasets must have the same type. Two datasets are of the same type if: • Both are defined as the same DATASET type • Both are defined as a subset of the same DATASET using filters • One is defined as a DATASET type, the other is a subset of the same type using filter Two datasets are not the same type if: • They are defined with different DATASET names, even if they have exactly the same RECORD structures. • One is defined as a DATASET type, the other is derived from the same DATASET type using TABLE(), even if they have exactly the same RECORD structure. The following example illustrates these rules. 01: aaa := DATASET(‘aaa’, {STRING1 fa; STRING1 fb;}, HOLE); 02: bbb := DATASET(‘bbb’, {STRING1 fa; STRING1 fb;}, HOLE); 03: 04: a1 := aaa(fa = ‘a’); 05: a2 := aaa(fa = ‘b’); 06: a3 := TABLE(aaa, {aaa.fa,aaa.fb}); 07: 08: OUTPUT(aaa+a1); // OK 09: OUTPUT(a1+a2); 10: 11: OUTPUT(a1+a3); // BAD 12: OUTPUT(aaa+bbb); In line 8, aaa is a dataset type, a1 is a subset of the same type using a filter (line 4). In line 10, both a1 and a2 are subsets of dataset type aaa. They are OK. In line 11, a3 is derived from aaa using TABLE(). In line 14, aaa and bbb are different dataset types, although they have exactly the same structure. Solution: Follow the above guidelines to make sure the two datasets involved are the same type. The fact that they have the same structure is not enough. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa; }, FLAT); 02: bbb := DATASET(‘bbb’, {STRING1 fa; }, HOLE); 03: 04: OUTPUT(aaa+bbb,{fa}); 05: OUTPUT(aaa-bbb,{fa}); 06: 07: a1 := aaa(fa = ‘a’); 08: a2 := aaa(fa = ‘b’); 09: 10: OUTPUT(a1+a2); 11: OUTPUT(a1-a2); 2013 The types must be the same The items that are in question must have the same type. More precisely, the types must be exactly the same, not just compatible. See Error C2012 for more details on how two dataset types are treated as the same type. Solution: Make sure they have the same type. Example: 01: aaa := DATASET(‘aaa’, {STRING1 a;}, HOLE); 02: bbb := DATASET(‘bbb’, {STRING1 b;}, HOLE); 03: ccc := aaa(a=’a’); 04: s1 := if(aaa=bbb,’true’,’false’); 05: s2 := if(aaa=ccc,’true’,’false’); 06: s3 := if(aaa!=bbb,’true’,’false’); 07: s4 := if(aaa!=ccc,’true’,’false’); In this example, we are comparing the equality of two datasets. Two dataset must be the same types if we want to compare them. 2014 List element has unknown type 2015 Integer type expected A valid INTEGER type is expected. A valid INTEGER type can be INTEGER or INTEGERn, where n is 1, 2, 4 or 8. Note that this error is different from Error C2002, where an integer value is expected. Solution: Make sure it is an valid INTEGER type. Example: 01: BIG_ENDIAN STRING i := ‘ABC’; 02: LITTLE_ENDIAN BOOLEAN b := FALSE; BIG_ENDIAN and LITTLE_ENDIAN can be used on valid INTEGER types only. 2016 Integer or decimal type expected A valid INTEGER or DECIMAL type is expected. A valid INTEGER type can be INTEGER or INTEGERn, where n is 1, 2, 4 or 8. A valid decimal type can be DECIMALn where n is an positive integer, or DECIMALn_y where n and y are positive integers. This normally happens after the UNSIGNED keyword, which can be applied to INTEGER or DECIMAL types only. Note that this error is different from Error C2001, 2002, where an integer value or real value is expected. Solution: Make sure it is a valid INTEGER or DECIMAL type. Example: UNSIGNED STRING s := ‘ABC’; UNSIGNED BOOLEAN b := FALSE; UNSIGNED cannot be used with STRING or BOOLEAN. 2017 String type expected A valid STRING type is expected. A valid STRING type can be STRING or STRINGn, where n is an integer that is greater than 0. The following list are the cases this error can occur: • ASCII type • EBCDIC type The bold type must be a valid STRING type. Note that this error is different from Error C2003, where a string value is expected. Solution: Change it to a valid STRING type. Example: 01: ASCII INTEGER s1 := 123; 02: EBCDIC BOOLEAN b := TRUE; All bold types should be a STRING type as expected by ASCII or EBCDIC. 2018 Indexing ALL is undefined 2019 Type mismatch - Integer or string value expected 2020 COUNT not valid in this context In the context where this error occurs, COUNT is not allowed. COUNT (as an alias to the COUNTER keyword) is only allowed in the following two cases: 1. As a parameter passed to the transform function in NORMALIZE: NORMALIZE(dataset, expression, myTranform (LEFT,COUNT)); 2. As a parameter passed to the transform function in DENORMALIZE: DENORMALIZE(dataset, expression, myTranform (LEFT,COUNT)); The COUNT keyword is not allowed to be used anywhere else. Please note that the COUNT() function can be applied to a dataset, as in COUNT(Person()); Solution: Make sure it is in the NORMALIZE or DENORMALIZE context. If you are using COUNT to count the number of tuples in a dataset, use the function COUNT(): the ( ) is required to use it as a function. Example: 01: NamesRec := RECORD 02: UNSIGNED1 numRows := 0; 03: STRING20 thename; 04: STRING20 addr1 := ‘’; 05: STRING20 addr2 := ‘’; 06: STRING20 addr3 := ‘’; 07: STRING20 addr4 := ‘’; 08: END; 09: 10: NamesTable := DATASET([ 11: {1, ‘Gavin’, ’10 Maltings Lane’}, 12: {2, ‘Liz’, ’10 Maltings Lane’, ‘3 The Cottages’}, 13: {0, ‘Mr Nobody’}, 14: {4, ‘Mr Everywhere’, ‘Here’, ‘There’, ‘Near’, ‘Far’} 15: ], NamesRec); 16: 17: OutRec := RECORD 18: UNSIGNED1 order; 19: UNSIGNED1 numRows; 20: STRING20 thename; 21: STRING20 addr; 22: END; 23: 24: OutRec NormalizeAddresses(NamesRec L, INTEGER C) := TRANSFORM 25: SELF := L; 26: SELF.order := C; 27: SELF.addr := CHOOSE(C, L.addr1, L.addr2, L.addr3, L.addr4); 28: END; 29: 30: NormalizeAddrs := NORMALIZE(namesTable, LEFT.numRows, 31: NormalizeAddresses(LEFT,COUNT)); 32: 33: OUTPUT(NormalizeAddrs); This example is a correct use of COUNT. We can also replace the COUNT keyword with COUNTER. The output of this example is: Order NumRows thename addr 1 1 Gavin 10 Maltings Lane 1 2 Liz 10 Maltings Lane 2 2 Liz 3 The Cottages 1 4 Mr Everywhere Here 2 4 Mr Everywhere There 3 4 Mr Everywhere Near 4 4 Mr Everywhere Far Example: MyValue := COUNT + 3; OUTPUT(Person( career = ‘Study’), {per_first_name, COUNT}); Both statements will produce a C2002 error since they are not in the context of NORMALIZE or DENORMALIZE. 2021 COUNTER not valid in this context The COUNTER keyword (as an alias of COUNT) is used only in NORMALIZE or DENORMALIZED. See Error 2020 for explanation, solution etc. 2022 LEFT not legal here The LEFT keyword is not legal in this context. LEFT can be used only in the following contexts: • Passing a parameter to a TRANSFORM function where a LEFT scope exists. • In TRIM(value, LEFT). LEFT means leading spaces. • In JOIN(dataset1, dataset2, join-fields, LEFT OUTER | LEFT ONLY). • In DEDUP(recordset, expression, LEFT). LEFT means to keep the head records from the comparison when duplication happens. Solution: Make sure LEFT is in one of the legal contexts listed above. Example: 01: a := EXP(LEFT); 02: NamesRec := RECORD 03: STRING20 thename; 04: END; 05: NamesTable := DATASET([{‘Gavin’}], NamesRec); 06: OutRec := RECORD 07: STRING20 thename; 08: END; 09: OutRec NormalizeAddresses(NamesRec L, INTEGER C) := 10: TRANSFORM 11: SELF := L; 12: END; 13: NormalizeAddrs := NORMALIZE(namesTable, LEFT.numRows, 14: NormalizeAddresses(LEFT,COUNT)); In line 1, LEFT is within none of the above contexts. In line 13 and 14, LEFT is used to specify the left record passed to the transform function. 2023 RIGHT not legal here This is very simliar to Error 2022. Refer to it for information. 2024 JOIN is not valid here See Error C2027. 2025 SELF not legal here SELF can only be used inside a TRANSFORM structure to indicate the current output record. Otherwise it is not legal. Solution: Don’t use SELF outside a TRANSFORM structure. 2026 ALL not legal here ALL is illegal in this context. ALL can only be used in the following contexts: • Inside a GROUP function • As a optional parameter to TRIM function • As a default value to a SET or SET OF parameter. Solution: Make sure it is one of the above cases, otherwise, don’t use ALL. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1;}, HOLE); 02: bbb := GROUP(aaa, ALL); 03: s := ‘ abc ‘; 04: TRIM(s, ALL); 05: f(SET is, INTEGER i) := i IN is; 06: MyJobSet := [1,2,3]; 07: BOOLEAN ok := f(ALL, 1); 08: SORT(aaa, ALL); All lines from 1-7 are legal. In line 8, ALL is not in any of the legal contexts and causes this error. 2027 JOINED can only be specified inside SORT or SORTED JOINED can only be used inside SORT or SORTED. Otherwise, it is illegal. Solution: Make sure it is in the SORT or SORTED. 2028 A value must be supplied for this attribute The attribute in question needs to be supplied a value. In ECL, we cannot just declare an at- tribute without defining its value. There is only one exception, that is when the attribute name starts with a $. Names that start with a $ will be used as environment variable in the future, but that functionality is not implemented yet. Solution: Use := to define a value to it. Example: 01: integer1 a; 02: EXPORT STRING1 s; 03: SHARED BOOLEAN b; Attributes a, s and b (whether it is local, EXPORTed, or SHARED) need a value when they are being defined. 2029 Constant doesn’t make any sense here Using a Constant doesn’t make sense in this context. For example, sorting or grouping by a constant doesn’t make sense. Example: 01: aaa := DATASET(‘aaa’,{STRING1 a; }, FLAT); 02: iftotal := IF(TRUE, 1, 0); 03: 04: countrec := RECORD 05: total := SUM(GROUP, iftotal); 06: END; 07: 08: a := OUTPUT(TABLE(aaa,countrec,1)); In line 8, the last 1 is not making sense. 2030 Illegal join flags for a lookup join 2031 DEDUP(file) not supported (or meaningful) on hole files 2032 Scheduling not allowed on an attribute (scheduling a definition makes no sense) 2033 Multiple workflow clauses 2034 Expected INDEX()... 2035 Type returned from transform must match the source dataset type 2036 Keyed parameter does not match the dataset passed as a second parameter 2037 Name of base file was not supplied and cannot be deduced 2038 BIAS not supplied and can't be calculated 2039 Character in string literal is not defined in encoding 2040 Invalid size for INTEGER type An INTEGER type can only be 1, 2, 4 or 8 bytes. The size is specified by appending the size right after the INTEGER keyword (no space allowed), as in: INTEGER1, INTEGER2, etc. By default, the size is 8; INTEGER is the same as INTEGER8. Solution: Make sure it is one of the following: INTEGER1, INTEGER2, INTEGER4, INTEGER8 or INTEGER. No space is allowed between INTEGER and the size digit. All identifiers that are INTEGER followed by a legal octal integer are reserved (for future extension, or just for the purpose of type checking). Therefore, INTEGER5 is neither a legal type, nor a legal identifier for attribute, function, formal parameter name, etc. Example: 01: INTEGER3 x := 10; 02: INTEGER10 y := 30; 03: f(INTEGER INTEGER32) := INTEGER32 + 3; In line 1 and 2, the size is wrong. In line 3, we can not use INTEGER32 as an identifier for formal parameter. 2041 Invalid size for REAL type A REAL type can be 4 or 8 bytes. So the valid forms are: REAL4, REAL8, and REAL. The last one will use the default size: 8. Solution: Choose the proper size. Example: REAL3 a := 1.2; 2042 Invalid size for DATA type The size of DATA type can be any positive integer; zero or negative integers are not allowed. When used as a function parameter type, the size can be omitted. Solution: Choose a proper size. Example: 01: DATA0 a := ’03DF21'; 2043 Invalid size for STRING type The size of STRING type can be any positive integer; zero or negative integers are not allowed. When used as a function parameter type, the size can be omitted. Solution: Choose a proper size. Example: STRING0 abc := ‘abc’; 2044 Invalid size for VARSTRING type The size of VARSTRING can be any positive integer; zero or negative integers are not allowed. Solution: Choose the proper size. Example: VARSTRING0 abc := ‘abc’; 2045 Invalid size for DECIMAL type A DECIMAL type does not specify or specifies an invalid size. An valid size must be specified for DECIMAL type. DECIMAL can be: • DECIMALn, where n can be 0-32 (inclusive) • DECIMALn_m, where n, m can be 0-32 (inclusive), and m must be less than or equal to n. In the above, n defines the number of digits in a packed decimal value while m defines the number of digits that appear after the decimal point. Solution: Choose a proper size. Example: 01: DECIMAL x1 := 20; 02: DECIMAL10_11 x2 := 20; In line 1, no size is defined. In line 2, the size of the fractional portion is greater than the total number of digits. 2046 A UDECIMAL type does not specify a size or specifies an invalid size. UDECIMAL is a short- hand alias for UNSIGNED DECIMAL. UDECIMAL can be: • UDECIMALn, where n can be 0-32 (inclusive), OR • UDECIMALn_m, where n, m can be 0-32 (inclusive), and m must be less than or equal to n. In the above, n defines the number of digits in a packed decimal value. m defines the number of digits after the decimal point. Solution: Choose a proper size. Example: 01: UDECIMAL x1 := 20; 02: UDECIMAL10_11 x2 := 20; 03: UDECIMAL34 x3 := 20; In line 1, no size is defined. In line 2, the size of percision is greater than the total length. In line 3, the size is greater than 32. Solution: Choose the proper size. 2047 Invalid size for BITFIELD type A BITFIELD (PROVISIONAL DOCUMENTATION, NOT YET IMPLEMENTED) type must have a valid size. A valid size for BITFIELD must: • explicitly specify the size • spcify the size in the range 1..64 (inclusive) Solution: Choose a proper size. Example: 01: MyRec := RECORD 02: BITFIELD sex := 1; 03: BITFIELD0 married := 1; 04: BITFIELD2 class := 1; 05: BITFIELD65 age := 100; 06: END; In line 2, sex does not specify a size. In line 3, married specifies a size of 0. In line 5, age speci- fies a size of 65. All these are not allowed. In line 4, class has size of 2, which is legal. 2048 Invalid size for UNSIGNED type UNSIGNEDn is a shorthand alias for UNSIGNED INTEGERn. So see Error 2040. 2049 Invalid size for QSTRING type 2050 Recursive macro call Recursive macro calls are not allowed. Recursive calls happen when a macro call itself directly or indirectly. Unlike functions, macros have no way to terminate once a recursive call starts; because the parser just does simple text substitution for macros. Recursive functions terminate at runtime by proper runtime logic. Note that since there are no forward references in ECL, it is also not possible to contruct a legal recursive function call. This situation may be changed in the future, however, recursive macros will still not be allowed due to the above reason. Solution: Change the code to eliminate the recursion. Example: 01: f(x) := MACRO 02: g(x)+3 03: ENDMACRO; 04: 05: g(x) := MACRO 06: f(x)+2 07: ENDMACRO; 08: 09: h(X) := MACRO 10: h(x)+1 11: ENDMACRO; 12: 13: abc := g(3); 14: xyz := h(4); In the example, macro f calls g, and g calls f again. Therefore we have an indirect recursive call. In macro h, it directly calls itself recursively. 2051 No matching ENDMACRO found The ENDMACRO keyword must terminate a MACRO definition. Due to the way macros are implemented internally, sometimes this error will appear in an unex- pected place. This happens within a macro call because, according to the grammar the parser thinks it should end the macro call, but the macro definition actually still has more code to generate. This normally indicates an error in the macro, the parameter(s) passed to the macro, or a combination of the two. Solution: Check the macro definition and the calls. Since the parser only realizes that an ENDMACRO is missing when it reaches the end of a file or a module, the position its report may be far away from the place it actually occurs. Example: 01: MyMacro(INTEGER i) := MACRO 02: i + 4; 03: 04: MyMacro(5); MyMacro is not terminated by an ENDMACRO. But the parser report the error at end of line 4 (the end of the file). 2052 EOF encountered while gathering macro parameters The file ends while the parser is still trying to look for macro parameters. This means that the code is not yet complete. Solution: Complete the code with proper parameters, and ending ‘)’. Example: 01: m2(x) := MACRO 02: 10+x 03: ENDMACRO; 04: m2(1 In line 4, macro call m2 doesn’t have complete parameters. 2053 No type is allowed for macro parameter 2054 Default value for macro parameter must be constant 2055 2056 2061 Too many parameters: XXX parameters expected Too many parameters are supplied for a function call or macro call. Solution: Check the definition of the function or macro. Supply the proper number of param- eters as required. Be careful about the scope rules in ECL, especially the local scope. See Ex- ample 2 below. Example 1: 01: f(x) := 4; 02: 03: m1() := MACRO 04: 10 05: ENDMACRO; 06: 07: m2(x) := MACRO 08: 10+x 09: ENDMACRO; 10: 11: f(2,3); 12: m1(1); 13: m2(1,2); All the calls in lines 11, 12, and 13 have extra parameters than that are needed. Example 2: 01: f(x) := x+1; 02: 03: EXPORT ab := 1; 04: 05: f(x,y) := x+y; 06: 07: f(3,4,1); In this example, function f is defined twice. Since an EXPORT or SHARED Attribute ends the local scope of an attribute/function/macro name, the f() in line 7 is a call to function f in line 5, not in line 1, therefore too many parameters are supplied. 2062 Omitted parameter has no default value A parameter to a function can be omittable when it is defined to have a default value. When a function is called, parameters that have default values can be omitted or not. However, those parameters that have no default values defined must pass a value for the parameter. Default value parameters in ECL are more general than in C++. In C++, if a parameter is defined to have default value, all the parameters following it must have default values defined as well, and if we supply a value to a parameter when we call a function, all parameters before this parameter must have values supplied. In ECL, these two constraints do not hold. The user can define default values for any set of parameters, regardless of their order in the parameter list. The user can also choose to pass values to those parameters that do have default values. See Example 2 below. Note that macros in ECL don’t have this default value concept, therefore this error can never happen to a macro call. Solution: Either pass a value for the parameter, or specify a default value in the function defini- tion, if possible. Example 1: 01: f(x,y,z=3) := x+y; 02: 03: f(12); 04: f(,2); In line 1, function f() only has a default value defined for the third parameter, so we must supply values for parameters 1 and 2. In line 3, no default value is given for parameter 2, and in line 4, no default value is given for parameter 1. Example 2: 01: f(x=1,y,z=3) := x+y; 02: 03: f(2,3,4); 04: f(,3); 05: f(,3,); 06: f(,3,4); 07: f(1,2); In line 1, function f() defines default values for parameters 1 and 3, which is perfectly OK in ECL, but is not allowed in C++. Calls in lines 3-7 are all legal in ECL, while calls in lines 4,5,6 are illegal in C++ (supposing line 1 were legal in C++). 2063 Too few parameters supplied Too few parameters are supplied to a macro call. Note that since function calls in ECL have the default value for omitted parameter rule (Error C2062), we can never get this error from a function call. Be careful with the scoping rules in ECL, as shown in Example 2. Example 1: 01: m(x) := MACRO 02: 10+x 03: ENDMACRO; 04: m(); 05: m2(x,y) := MACRO 06: 10+x 07: ENDMACRO; 08: m2(1); The macro calls in both lines 4 and 8 are supplied fewer parameters than required. Example 2: 01: m() := MACRO 02: 2 03: ENDMACRO; 04: SHARED y:= 3; 05: m(x) := MACRO 06: x+2 07: ENDMACRO; 08: m(); In this example, call m() in line 8 is to the macro defined in lines 5-7, not the macro defined in lines 1-3. The reason is that the SHARED statement in line 4 ends the scope of local macro m() in line 1. 2064 param type mismatch The parameter supplied to a function call has a different type than the type specified for the function definition. Automatic type promotion (implicit type casting) will be applied when the compiler tries to match the types. Therefore, we can legally use an integer to call a function that requires a real parameter. Narrow casting (down casting from real to integer) is also applied when proper. A macro is used for text substitution, only. There is no type information associated with a macro parameter. Therefore, this error only applies to function calls. Solution: Check the function definition and make sure the types match. Use type cast when necessary. Example: 01: f(integer i) := i+1; 02: f(‘123’); In line 2, actual parameter ‘123’ is a string, not an integer as expected by the definition of f. We can use type casting to correct this: f((INTEGER)‘123’); 2065 Wrong number of parameters: either 2061 or 2062 2066 non-typed or void expression can not used as parameter 2071 Constant expression expected A constant expression is expected. A constant expression can be: any constant value, such as: 123, ‘abc’. any expression that evaluates to a constant value, such as: 2*4+5, ‘abc’+’def ’, 3>2 && 3^2=8. any function or macro that evaluates at compile time to a constant value. For example, if we have the following: c(x) := x*2; y := 3;Then c(y) is a constant since at compile time, it can be evaluted as a constant value 6. The third rule is different from languages like C and Java where a function’s return value is never regarded as a compile time constant, macros in C (Java doesn’t have macros) are not an issue since they disappear after preprocessing. The following are not constants: RANDOM(), or any expression that contains RANDOM() The result from a dataset, such as COUNT(Person()) Any expression that contains any non-constant value(s). Solution: Choose another name for the field, or remove one if they are just duplicated. Example 1: 01: STRING8 MyString1 := INTFORMAT(10, RANDOM(), 1); 02: c(x) := x*2; 03: y := 3; 04: STRING8 MyString2 := INTFORMAT(10, c(y), 1); 05: ct := COUNT(Person()); 06: STRING8 MyString3 := INTFORMAT(10, ct, 1); Both RANDOM() in line 1 and ct in line 6 are not constant, but c(y) in line 4 is constant. Example 2: 01: loadxml(‘myxml.xml’); 02: #DECLARE (S) 03: #SET (S, RANDOM()) 04: #APPEND(S, RANDOM()) 05: #IF (RANDOM()) 06: #APPEND(S, ‘x := 1; ‘) 07: #ELSE 08: #APPEND(S, ‘x := 2; ‘) 09: #END All these RANDOM() function calls are illegal since constants are expected where they appear. To test this, an xml file named myxml.xml must be provided, but the content of the file is not important for our concern. 2072 Expected numeric expression An expression that is not compatible with a numeric type is encountered. A numeric expression can be: INTEGER, REAL, or DECIMAL (for all legal size, and signed, unsigned types). This error happens in the +, -, *, / operations. In the expression: expr1 + expr2 if one of the expr1 and expr2 is string, the other is implicitly cast to string (if it is not already string, and can be cast to string). For instance: x := 123 + ‘STR’ is a legal expression, and x has the value: ‘123STR’. For other operations (-, *, /), only numeric values are allowed. Implicit cast may be involved, but: • A string is not implicitly cast to a numeric value (while an numeric can be implicitly cast to a string value). • A boolean vaue cannot be cast to either string or numeric values. • A DATA value can only be cast to or from STRING, and not to or from a numeric value. Solution: Change the expression to be a numeric type. Use explict casting where proper. Example: 01: x1 := TRUE + 35; 02: 03: x2 := 35 - ‘STR’; 04: 05: x3 := 34 * ‘STR’; 06: 07: x4 := 12 * x’abcd’; All bold expressions are not numeric (and cannot be implicitly cast to a numeric value). 2073 Expected boolean expression A boolean expression is expected where this error occurs. For C programmers, integer is not compatible with boolean in ECL. So 0, 1 or 2 are not a legal boolean values as in C. You can cast an INTEGER to a BOOLEAN. The following is a partial list of where a boolean expression is expected: • IF(condition, expr1,expr2) • MAP(condtion1=>expr1,…,condition_n=>exprn, defvalue) • IFBLOCK(condition) • WHEN • WHICH(condition1,…,conditions) • REJECTED(condition1,…,conditions) • #IF(condition) Solution: Use boolean expression instead. Use comparison (=, !=, <>, etc.) to change an integer to a boolean when proper. Example: 01: x1 := NOT 2; 02: x2 := FALSE AND ‘STR’; 03: x3 := TRUE OR x’abcd’; 04: x4 := IF (2,3,4); 05: x5 := MAP(1=>3,4); 06: x6 := WHICH(‘2’,3<4); All the bol expressions should be replaced by boolean expressions. 2074 ‘(‘ expected An open parenthesis ‘(’ is missing. This error normally happens in ECL template commands: • #DECLARE • #IF • #FOR • #SET Solution: Add a ‘(‘ and ‘)’ if needed. Example: 01: loadxml(‘myxml.xml’) 02: 03: #DECLARE s 04: 05: #IF TRUE 06: #END 07: 08: #SET s 09: 10: #FOR item(%’type’% = ‘count’) 11: #APPEND(s, ‘count’) 12: #END All bold expressions should be surrounded by parentheses. 2075 ‘)’ expected A close parenthesis ‘)’ is missing. This error normally happens in ECL template commands: Solution: Add the missing ‘)’. Example: 01: loadxml(‘myxml.xml’) 02: 03: #DECLARE (s 04: 05: #IF (TRUE 06: #END 07: 08: #SET (s,3 09: 10: #FOR (item(%’type’% = ‘cunt’) 11: #APPEND(s, ‘count’) 12: #END The compiler may not do well enough to point out every missing ‘)’ in this example, so looking around is always a good idea. 2076 ',' expected A comma (‘,’) is expected. Normally this means another parameter is expected, or the program construct is incomplete. Solution: Check the context where the error occurs and make sure to supply complete parameters or complete constructs. Example: 01: #SET (s) #SET expects two parameters instead of one. 2077 identifier expected An identifier is expected. This normally means the code expected an attribute name. Solution: Check the context, and make sure understand the construct you are using. Example: 01: #DECLARE (1) 02: #SET(1,2) 2078 ; expected 2079 Expected an 8 byte file position 2080 Fileposition should be the last field in an index 2081 Import names unknown module XXX The IMPORT names a module that appears not to exist. In ECL, ll modules are defined and stored in a central ECL repository. Solution: Make sure the module exists in the ECL repository and its name is spelled correctly for the IMPORT. If you believe the module is there and you still have problem, this is often a system configuration problem. Unless you are unfamiliar with the system, contact your system administrator. The following information is only for the advanced user who will do the configuration by him/ herself. If you are using Query Builder program, use the Connections tab in the File/Preferences dialog. Use the Quick Config button if you can. If youare running a local ECL server, you can first use the Quick Config to set up the Jack and Hole cluster. Then change the ECL server to point to you machine (either the machine name or IP will be OK). Example: IMPORT unknown_module; In the above example, unknown_module is not the ECL repository. 2091 JOINED must specify a sorted dataset In the SORT function: SORT (recordset, value [, value …], JOINED(joinedset)]); the parameter to JOINED must be a sorted set. A sorted dataset is either a dataset that is labeled as SORTED, or the result set of a SORT function. Solution: Label the dataset as SORTED if it is already sorted, otherwise sort it first. Normally, we need a temporary attribute to hold the sorted dataset because of Error C2161. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa}, hole); 02: bbb := DATASET(‘bbb’, {STRING1 fb}, hole, aaa); 03: sorted_bbb := SORTED(bbb, fb); 04: OUTPUT(SORT(aaa, fa, JOINED(bbb))); The bbb is not a sorted dataset in line 4, so we get this error. sorted_bbb is sorted, so we can solve this problm by replacing bbb with sorted_bbb. 2092 JOINED data set has different number of sort fields In the SORT statement: SORT (recordset, value, [value…], JOINED(joinedset)); the number of sort fields in joinedset and the number of field to be sorted on for the recordset must be the same. The sort fields are the fields that are specified in the SORT or SORTED statement. For example: SortedPerson := SORT(person, first_name, last_name); has 2 sort fields: first_name, and last_name, and sortedPerson2 := SORTED(sortedPerson, first_name); has only 1 sort field: first_name as specified in the SORTED statement, although sortedPerson was sorted on two fields. Solution: Specify the same number of sort fields as required. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1, STRING1 f2}, hole); 02: sored_aaa := SORT(aaa,f1,f2); 03: OUTPUT(SORT(aaa, f2, JOINED(sorted_aaa))); 04: // this is OK although sorted_aaa was sorted on 2 fields. 05: sorted_aaa2 := SORTED(sorted_aaa,f1); 06: OUTPUT(SORT(aaa, f2, JOINED(sorted_aaa2))); 07: bbb := DATASET(‘bbb’, {STRING1 fb}, hole); 08: sorted_bbb := SORTED(bbb,fb); 09: OUTPUT(SORT(aaa, f1, f2, JOINED(sorted_bbb))); In line 3, sorted_aaa has 2 sort fields (f1, f2 as defined in line 2), while the outer sort to aaa only has 1 field (f2). In line 9, sorted_bbb has 1 sort field while sort on aaa has 2. Both cases are not allowed. Note that line 6 is OK since sorted_aaa2 has only 1 sort field specified although sorted_aaa was sorted on two fields. 2093 Component of JOINED has different type to this sort In the SORT statement: SORT (recordset, value, [value…], JOINED(joinedset)); the type of each field specified in SORT (by value, [value…]) must be the same as the sort fields in the joinedset. The type is compared in the order they are specified. (Note that they must have the same number of fields, see Error 2092.) Solution: Make sure they have the same types. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1, BOOLEAN f2}, hole); 02: sorted_aaa := SORT(aaa,f1); 03: OUTPUT(SORT(aaa, f2, JOINED(sorted_aaa))); 04: 05: sorted_aaa1 := SORT(aaa,f1,f2); 06: OUTPUT(SORT(aaa, f2, f1, JOINED(sorted_aaa1))); In line 2, sorted_aaa is sorted on f1, whose type is STRING1. However, in line 3, we are sorting aa on field f2 which is of type BOOLEAN. In line 6, we should know that the order of the fields matters. 2094 Too many joined clauses In a SORT statement, only one JOINED clause is allowed. If we give more than one JOINED clause, we get this error. Solution: Remove extra JOINED clause, and only keep one. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1, BOOLEAN f2}, hole); 02: sorted_aaa := SORT(aaa,f1); 03: OUTPUT(SORT(aaa, f1, JOINED(sorted_aaa), JOINED(sorted_aaa))); Two JOINED clauses are presented in the SORT of line 3. 2095 Implicit joins are only supported in HOLE 2096 The list to be sorted on can not be empty 2097 Invalid size for UNICODE or VARUNICODE type 2098 Incompatible locales in unicode arguments of binary operation 2099 Bad locale name 2100 Definition must define EXPORTed or SHAREed value for XXX. You are trying to use an attribute that is defined in a module but not EXPORTed or SHARED. If you want use an attribute that is defined in another module, the attribute must be EXPORTed. If you are using an attribute defined in the same module, bt in a different file (a Query Build window), it must be defined as SHARED. Solution: Export or share the attribute. Example: Inside MyModule: 01: SmallPersonSet := CHOOSEN(Person(), 50); ECL code that is trying to use SmallPersonSet: 01: import MyModule; 02: output(MyModule.SmallPersonSet); We get this error since SmallPersonSet is defined, but neither EXPORTed or SHARED. 2101 Error in referenced attribute 2110 A value for XXX has already been specified In a TRANSFORM definition, each field of the return record structure must receive a value once (otherwise Error 2111 occurs) and only once (otherwise this error occurs). Note that the collective assignment definition: SELF := label; places a value in each field of the result record from its matching field in the label record. How- ever, this collective assignment definition can be overwritten by a following individual assign- ment definition to one or more of the fields. This is a special exception. The execution engine can detect this situation if a following assignment definition exists. Solution: Make sure to only define each output field once. Use CSE, MAP, CHOOSE, or IF when proper so that there is only one assignment definition for each output ield. Example: 01: NamesRec := RECORD 02: STRING20 thename; 03: STRING20 addr1 := ‘’; 04: STRING20 addr2 := ‘’; 05: END; 06: 07: OutRec := RECORD 08: STRING20 thename; 09: STRING20 addr; 10: END; 11: 12: OutRec Trans(NamesRec L, INTEGER C) := 13: TRANSFORM 14: SELF := L; 15: SELF.addr := CHOOSE(C, L.addr1, L.addr2); 16: SELF.addr := ‘ABC’; 17: END; 2111 Transform does not supply a value for field XXX In a TRANSFORM definition, each field of the return record structure must recieve a value, either explicitly by an assignment definition, or implicitly by SELF := label; Either way, we must guarantee each field receives a value. Solution: Ensure a value is placed in the field. If you want the field just get the value of a corre- sponding filed in an input record, the two fields must have the same names. Example: 01: NamesRec := RECORD 02: //STRING20 thename; 03: STRING20 addr1 := ‘’; 04: STRING20 addr2 := ‘’; 05: END; 06: 07: OutRec := RECORD 08: STRING20 thename; 09: STRING20 addr; 10: END; 11: 12: OutRec Trans(NamesRec L, INTEGER C) := TRANFORM 13: SELF := L; 14: SELF.addr := CHOOSE(C, L.addr1, L.addr2); 15: END; Since we commented out the field thename in NamesRec definition (line 2), and we don’t explicitly place a value in OutRec’s thename field (line 13-15), we get this error. We can either un- comment line 2, or place a value in OutRec.thename within the transform to resolve the problem. 2112 A field called XXX is already defined in this record A field with the same name as the one in question is already defined in the record. Field names must be unique within a record definition. It doesn’t matter if their types are the same or not. Solution: Choose another name for the field, or remove one if they are just duplicated. Example: 01: MyRec := RECORD 02: STRING20 name; 03: INTEGER1 age; 04: STRING20 name; 05: INTEGER1 name; 06: END; In the above example, name is aleady defined in line 2, the definitions in line 4 and 5 are not allowed. 2113 TRANSFORM required a record return type 2114 Can not assign type XXX to self 2115 Transform must have at least 1 parameter 2116 Dataset can not be used in transform 2117 Can not use dataset directly in transform parameters 2121 Invalid substring range The index to a string is in an illegal range. There are 3 cases: • If a start index is specified, it must e in the range [1 .. string_len], inclusive. • If a stop index is specified, it must be in the range [1 .. string_len], inclusive. • If both start index and stop indexes are specified, the start index must be less than or equal to the stop index. The <more info> will explain which case the error is. If there are mutiple cases in one substring expression, only the first error is given (to reduce the number of error messages). For the C/ C++/Java programmers, ECL indexing always starts at element 1 (not 0). This error is a compiler error, not a runtime error. Don’t count on the compiler to do all the index checking for you. It is always a good idea to mae sure the index is in proper range. For example: 01: STRING concat(STRING x, STRING y) := x+y; 02: STRING a := ‘a’; 03: STRING b := ‘bc’; 04: STRING c := concat(a,b); 05: c[10]; The compiler cannot know that the index value (10) in line 5 is out of range. The behavior is undefined if the index is out of range at runtime. Solution: For indexing string constants, make sure it is in proper range. For strings of unknown size (like a STRING parameter passed to a function), use the LENGTH function as necessary. Example: 01: STRING3 s := ‘abc’; 02: s1 := s[0]; 03: s2 := s[1..4]; 04: s3 := s[3..2]; We get the following error message: error C2121: Invalid substring range: start index out of [1..len] error C2121: Invalid substring range: end index out of [1..len] error C2121: Invalid substring range: start index > end index 2124 STORED() has zero length 2130 Value for field XXX cannot be computed in this scope Inside an OUTPUT, the output record specifies a field that is not part of the dataset. The example will help to clarify this. Solution: Remove the offending field from the output record set. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa }, HOLE); 02: bbb := DATASET(‘bbb’, {STRING1 fb }, HOLE); 03: 04: OUTPUT(aaa, {aaa.fa, bbb.fb}); In line 4, field bbb.fb is not a field of output dataset aaa, therefore cannot be specified here. 2131 Incorrect assertion scoping There is a problem with the scoping of the referenced attribute. There are several cases. • In a filter, each referenced field must be a field of the dataset we are filtering. The examplewill help to clarify this. • In SORT, the field to be sorted must be a field of the dataset that is being sorted. • In SORTED, the field that is specified as having been sorted on must be a field of the dataset. And much more… Solution: Remove the offending field from the filter. Example 1: 01: aaa := DATASET(‘aaa’, {STRING1 fa }, HOLE); 02: bbb := DATASET(‘bbb’, {STRING1 fb }, HOLE); 03: 04: OUTPUT(aaa(fa=’Good’, bbb.fb=’Bad’)); This is a filter example. In line 4, the filter bbb.fb = ‘Bad’ refers to a field of dataset bbb, but we are trying to filter dataset aaa. Example 2: 01: aaa := DATASET(‘aaa’, {STRING1 fa }, HOLE); 02: bbb := DATASET(‘aaa’, {STRING1 fb }, HOLE); 03: ccc := SORT(aaa, fa, bbb.fb); This is a SORT example. In line 3, bbb.fb is not a field of the dataset aaa which we are sorting. 2132 Assertion must be boolean Each filter for a dataset must be a boolean expression. This error usually follows error C2073: Boolean expression expected. Solution: Change the expression in question to a boolean expression. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1; STRING1 f2; }, HOLE); 02: bbb := aaa(f1); 03: ccc := aaa(f1*2); In line 2, f1 is not boolean type, and in line 3, f1*2 gives an error, and results in an unknown type filter. 2133 Use dataset in expression without proper context 2134 Parameter to fetch isn't a dataset 2141 Illegal combination of modifers The modifiers to attributes have an illegal combination. Duplicating the same modifier is an example of illegal combination. Combining some conflicted modifiers will be an error too, but will cause different errors than this one (e.g., see Error C2142). Solution: Remove any duplicated modifier. Example: 01: SHARED SHARED aa := 3; Here two SHARED are used at the same time – an illegal combination. Remove one of them. 2142 EXPORT and SHARED cannot be specified together EXPORT and SHARED are both specified on an attribute, which is not allowed. Solution: Remove EXPORT or SHARED. Example: 01: SHARED EXPORT aa := 3; 02: EXPORT SHARED bb := 3; Both lines specify SHARED and EXPORT at the same time. 2143 Identifier XXX is already defined The identifier in question is already defined somewhere before. The identifier can be an attribute name, a user-defined function name, a user-defined macro name, etc. Functions, macros, and attributes all share a name space. Therefore, an identifier that is used as a function name cannot be used again in the same scope. See Scope rules in ECL for information about scoping. In ECL, an attribute can be defined only once in a given scope. Once defined, we cannot rede- fine it again in the same scope. In this regard, attributes are different from variables in the normal procedure programming languages. Solution: Choose another name, or make the conflicting names exist in different scopes, if possible. Removing EXPORT or SHARED to changethe scope level may make once-valid code invalid since EXPORT or SHARED changes the scope of any local attributes before it. Example: 01: attr := 4; 02: attr := 3; 03: 04: func(x) := x*2; 05: func(x) := x+2; 06: 07: macro1 := MACRO 08: 4 09: ENDMACRO; 10: macro1 := 2; In the example, all identifiers in bold are already defined before it. 2144 Expression produces string field of unknown length The compiler cannot determine the size of a string field. This normally is the result of some other ECL error that causes the compiler to have trouble evaluating the size of a string. This error happens often in RECORD definitions (and its IFBLOCK). Solution: If the error happens in a RECORD definition, make sure the field has a name, and a STRINGn type (STRING without a length cannot be used, otherwise you get a C2043 error). The field can be anonymous only if: (1) the data type is omitted, and (2) a dataset name is used as the expression (which is used to include all the fields from the dataset at their declared sizes and value types). For example: aaa := DATASET(‘aaa’,{STRING1 f1, INTEGER1 f2}, FLAT); MyRec := RECORD aaa; STRING1 sex; END; This is equal to: MyRec := RECORD STRING1 f1; INTEGER1 f2; STRING1 sex; END; Example: 01: STRING cat(STRING s1, STRING s2) := s1 + s2; 02: 03: MyRec := RECORD 04: cat(‘abc’,’def’); 05: END; In line 4, no name is given to the to-be-defined field of MyRec. The compiler tries to decide the type of the expression cat(‘abc’, ‘def ’) and discovers it is STRING (with no specific length) from the definition the cat function in line 1. Therefore you get this unknown size error. Giving a name and a specific size type to the field solves the problem: 04: STRING6 magictag := cat(‘abc’,’def’); Note that we also have to give a size explicitly to the STRING type in the above code, otherwise, an error C2043 occurs. 2145 Different character sets in concatenation The operands in string concatenation (+) have different character sets. In ECL, we can use ASCII or EBCDIC charset. By default, a STRING is ASCII. We can not concatenate an EB- CDIC string to an ASCII string (or vice versa). Solution: Concatenate an EBCDIC string and an ASCII string seldom make sense. If you do want to do this, what you really want to do is to change the code of one string so that the two strings have the same charsets. An ECL function or service can be defined to accomplish this task. There is no build in function or service to do this (AFAIK). Example: 01: ASCII STRING3 x := ‘ABC’; 02: EBCDIC STRING4 y := ‘ABCD’; 03: STRING7 a := x + y; Note that in line 3, y is not an ASCII string ‘ABCD’ since we are defining it as an EBCDIC encoded string. 2146 Type transfer: target type is larger than source type In a type transfer, the size of target type is larger the size of source type. ECL doesn’t allow this because it doesn’t know how to fill the extra space in the target. However, it quietly accepts the other case: target size is smaller than the source type in which case data is lost. It assumes that the user doing the transfer knows what he is doing. Type transfer in ECL can be done using the build-in function TRANSFER or use the transfer operator: (>, <). They are equivalent. Solution: Make sure the size of the target type is equal to or less than that of the source type. Example: 01: INTEGER2 x := 65; 02: STRING3 xstr1 := TRANSFER(x,STRING3); 03: STRING3 xstr2 := (>STRING3<)x; In line 1, x is defined as INTEGER2 (which has size 2). In line 2 and 3, both target types are STRING3 (size=3). Line 2 and line 3 are doing the same thing: one uses TRANSFER function, the other uses (>, <) operator. 2147 Illegal identifier XXX: $ is not allowed 2160 SIZEOF requires a field or dataset parameter The build-in function SIZEOF only works with a field or DATASET. For C programmers, SIZEOF in ECL is different from the sizeof operator in C/C++, where it can be used on a data type or a variable. Solution: Use SIZEOF only on a field or a dataset as required. Example: 01: aaa := DATASET(‘aaa’, {STRING1 f1, STRING1 f2}, hole); 02: OUTPUT(aaa, {f1,SIZEOF(aaa),SIZEOF(f1)}); 03: 04: INTEGER4 x := 3; 05: sizeOfInt := SIZEOF(x); 06: sizeOfInt4 := SIZEOF(INTEGER4); In line 2, SIZEOF is on dataset aaa or field f1. Therefore, it is OK. However, in lines 5 and 6, SIZEOF is on an attribute x and a data type INTEGER4, which will causing this compile error. 2161 Nested SORTs may behave incorrectly. Split the attribute into two Sort function calls are nested and may behave incorrectly. This is a limit of the current implementation. In the future, this limitation may be removed. Solution: Split the attribute into two. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa}, hole); 02: bbb := DATASET(‘bbb’, {STRING1 fb}, hole, aaa); 03: OUTPUT(SORT(aaa, SORT(bbb, fb)[1].fb)); The compiler complains on the second SORT in line 3. We can solve the problem by splitting the attribute, like this: 01: aaa := DATASET(‘aaa’, {STRING1 fa}, hole); 02: bbb := DATASET(‘bbb’, {STRING1 fb}, hole, aaa); 03: temp := SORT(bbb, fb); 04: OUTPUT(SORT(aaa, temp[1].fb)); 2162 Nested GROUPs may behave incorrectly. Split the attribute into two Group function calls are nested and may behave incorrectly. This is a limit of the current implementation. In the future, this limitation may be removed. Solution: Split the attribute into two. Example: 01: aaa := DATASET(‘aaa’, {STRING1 fa}, hole); 02: bbb := DATASET(‘bbb’, {STRING1 fb}, hole, aaa); 03: OUTPUT(GROUP(aaa, GROUP(bbb, fb)[1].fb)); The compiler complains on the second GROUP in line 3. We can solve the problem by splitting the attribute, like this: 01: aaa := DATASET(‘aaa’, {STRING1 fa}, hole); 02: bbb := DATASET(‘bbb’, {STRING1 fb}, hole, aaa); 03: temp := GROUP(bbb, fb); 04: OUTPUT(GROUP(aaa, temp[1].fb)); 2163 Too many operands This error happens in the ECL template commands. The command expected a certain number of parameters, and the code supplies more than it needs. Solution: Reduce the number of parameters as required. Example: 01: LOADXML(‘myxml.xml’); 02: 03: #DECLARE(S) 04: 05: #IF(S > 1, S < 5) 06: #ENDIF 07: 08: #SET (S, ‘1’, 2) Both #IF and #SET need fewer parameters than they are supplied. 2164 No XML scope active The current statement is legal only in an XML context. For example, the template language is designed specifically for parsing XML code, and therefore requires an active XML source. Solution: Add an XML scope before the error (normally at the beginning of the file) using LOADXML(xml-file-name). Example: 01: #DECLARE(s) We can simply add a dummy loadxml statement before line 1 to resolve this error: LOADXML(“myxml.xml”); The content for this example is not important, but it is important that the file does exist. 2165 File extension must be ".HQL" The file extension used to provide input for an application is illegal. For example, Seisint’s internal test program hqltest requires the .hql file extension on its input files. Solution: Change the file extension to the one required for the specific application. Example: N/A. 2166 File XXX not found A file is not found. This only happens with Seisint’s internal test programs hqltest or ecltest. Solution: Check the file name, and make sure the file exists in the proper directory. If the path is relative, it is relative to the current (working) directory. 2167 Unknown identifier The identifer is unknown for one of the following reasons: • The identifier has been misspelled. • The identifier names an attribute, function, dataset, or macro defined in another module that has not been imported. • The identifier is defined but it is not visible due to the scoping rules of ECL. In particu- lar, the identifier is probably local (not SHARED or EXPORTed), and there is an EX- PORT or SHARED attribute defintion that terminates its scope. • It is a field of a record, a dataset etc, and it is not properly qualified. Solution: First make sure it is spelled correctly. If it is defined in other module, make sure it is properly imported and qualified. If it is a field of some structure (like record, or table), qualify it properly (such as: myrecord.field, mytable.field). If it is defined as a local attribute to another SHARED and EXPORT attribute, either promote it to SHARED or EXPORTed scope, or move the code around so that the identifier’s defintion and the defintion using it are within the same local scope. Example: 01: a := f(x); 02: f(x) := x+3; 03: EXPORT y := 4; 04: a := f(y); In line 1, f is not defined in this module. In line 4, f is alreaday defined in line 2, but the scope of f in line 2 is terminated by the EXPORT in line 3. Make f in line 2 SHARED or move line 2 after line 3 to solve the problem. 2168 Select a field other than these grouped upon 2169 Aggregate field following a field that has a variable length 2170 No default value for a field A default value is required for some fields in a record. See example for additional information. Solution: Given a default value if proper. Most likely, a default value is not what you need al- though it may solve the compiler error problem. Example: 01: bbb := DATASET(‘aaa’, {STRING1 fa; }, FLAT); 02: table1 := TABLE(bbb, {STRING1 fa; STRING1 fb; }); In this example, we create a table from dataset aaa. Since bbb has a field called fa, we can have a value for field fa in the table. However, there is no such field called fb in dataset bbb. Hence we get an error saying that Field fb does not have default value. In this case, assign a default value to fb. 2171 Object has no such a field 2180 Unknown escape sequence Astring constant contains an unknown escape sequence. Currently, the only supported escape sequences are: \’ and \\ For C/C++/Java programmers, don’t assume that ECL supports the standard escape sequences in these languages. Solution: Remove the offending escape sequence. Example: 01: string x1 := ‘\r\n\t\b\x0123’; None of the above C escape sequences is supported in ECL. 2181 A string end with escape char \, eg., x := 'abc\'; A string constant is terminated by a backslash (\). Most of the time, the intention of the user is to have a single quote (‘) as the last character in the string. As a result, he/she puts an escape character (\) in front of the single quote (‘) indicating the end of the string constant and the terminating quote is ignored. Solution: Add another ‘ to terminate the string or remove the offending \ whichever is proper. Example: 01: String s := ‘It is brothers\’; The last character in this string is ‘, which has to be escaped as \’, therefore we need another ‘ to terminate the string constant. 2182 Illegal hex data, it can only have 0-9a-fA-F Illegal characters are present in hex data. Hex data is defined as: X’hex-digits’ or x’hex-digits’ A hex digit may only be any of the numbers 0 through 9 or the letters A through F (upper or lower case). The number of digits in the hex string must be even (see Warning C1002) and be greater than 0 (see Warning C1003). Solution: Remove the illegal characters in the hex. Example: 01: DATA x1 := x’ABCX’; 02: DATA x2 := x’^1'; 03: DATA x3 := x’A B’; 04: DATA x4 := x’\n’; All bold chars are illegal hex digits (line 3 has a blank space between A and B), therefore none of the above are legal hex data strings. 2183 Odd number of digits in hex data, e.g., x'ABC'. In a hex data constant, the number of digits in the string must be even because it takes two hex digits to define the value of a single byte of data. Solution: Check that there are no missing or extra digits. Add a leading or trailing 0 as required. Example: DATA d := x’ABCD123'; The hex data string x’ABCD123' contains 7 digits (an odd number). Make it: x’0ABCD123'. 2184 Empty hex data, e.g., "x". The hex data cannot be empty. Solution: Change it to meaningful, non-empty hex data. Example: DATA d = ‘’; 2185 Comment is not started: ending is not allowed 2186 Char XXX needs to be escaped 2187 #ERROR statement 2190 Empty record definition: no field defined In a record definition, no field is defined. This is not allowed in ECL. For C/C++/Java programmers where empty structures (struct, class etc) are allowed, note that this is not allowed in ECL. Solution: Provide at least one field for the record. If you really don’t need a real field, add a dummy field. Example: 01: MyRec := RECORD 02: END 2191 Empty substring range The sub string expression is empty. In ECL, a sub string can be: • str [start_index .. end_index] • str [ .. end_index] • str [ start_index .. ] • str [ one_index ] Each index is an integer within the range of 1…LENGTH(str), inclusive. Solution: Either provide an index value or remove the square brackets. Example: 01: STRING s := ‘abc’; 02: STRING subs := s[]; 2192 Empty user type definition The user type definition body is empty. In ECL, user types are defined by the TYPE structure. Solution: Provide at least one function for the defintion. Example: 01: MyType := TYPE 02: END 2193 Empty transform definition The TRANSFORM definition body is empty. Since a RECORD type can not be empty (see error C2190), and each record must be assigned a value in the transform (see error C2111), an empty transform definition cannot be legal. Solution: Provide at least a trival SELF assignment. Example: 01: NamesRec := RECORD 02: STRING20 thename; 03: END; 04: 05: OutRec := RECORD 06: STRING20 thename; 07: END; 08: 09: OutRec Trans(NamesRec L, INTEGER C) := 10: TRANSFORM 11: END; Adding the following trival assignment definition in the TRANSFORM will solve the problem: SELF := L; 2194 Unended comment This error indicates that EOF was reached while a comment is not ended. This only occurs to C- style comment: /* and */. Solution: Add an ending */ in the proper position. Example: 01: a := 1; 02: /* this is a unended comment .. The ending */ is missing in the above example. 2195 Unended string constant: string must end in one line In ECL, a string constant must be complete in one line. The string delimitor is single quote (‘). Note that ECL doesn’t support string constants that span multiple lines. However, you can use the concatenation operator (+) to reach the goal. Since constants are folded at compile time, there is no performance penalty for this. Solution: Add a single quote at the end to terminate the string. Example: 01: STRING long_string := ‘the part on first line’ 02: + ’the part on second line ‘ 03: + ‘and the third line’; 04: 05: STRING s := ‘unended string const; In line 1-3, we are defining a long string which spans multiple lines. The string in line 5 doesn’t end with a single quote. Note that the ‘;’ is regarded as part of the string, and will not end the string constant. 2196 is illegal string delimiter: use ' instead In ECL, the string constant delimiter is the single quote (‘) character, not double quotes as in some other languages (C/C++/Java etc). Since single quote is the delimiter, it needs to be escaped if it is part of the string, like: ‘He\’s a nice guy’. And double quote does not need (and cannot use) escape. Solution: Change double quote to single quote Example: 01: string s := “abc”; 2197 Empty ifblock definition 2200 EOF encountered inside #FOR The File ends while we are examining a #FOR command. This usually means the terminating #END is missing. Solution: Add #END if proper. Example: 01: loadxml(‘myxml.xml’); 02: #DECLARE (x) 03: #SET(x, 1) 04: 05: #FOR (x) 06: #APPEND(s, ‘1’) 2201 EOF inside parameter gathering EOF is encountered when the parser is trying to gather parameters. This normally occurs in a template command. Solution: Make sure the ECL is complete. Example: 01: LOADXML(‘<xml>dummy</xml>’); 02: #IF( The #IF doesn’t have a parameter yet. 2202 Template symbol has already been declared In ECL template language, a symbol can only be declared once. You get this error if you are trying to declare it a second time. Pay attention to the scope rule. A symbol can be defined multiple times if it belongs to different scope each time. However, if you change the scoping, they may become within the same scope and therefore generate this error. Solution: Make sure to only declare each symbol once in the current scope. Example: 01: LOADXML(‘<xml>x</xml>’); 02: 03: #DECLARE(s) 04: 05: #FOR(i) 06: #DECLARE(s) 07: #BREAK 08: #END 09: 10: #DECLARE(s) Since #FOR will start a new local scope, the #DECLARE in line 6 doesn’t cause problem. However, line 10 does cause problem, because s is already defined in line 3 (which is in the same scope). 2203 Template symbol has not been declared In ECL template language, a symbol must be declared before it can be used. A symbol is de- clared using the #DECLARE command. Solution: Make sure you declare it before you use it. Example: 01: LOADXML(‘<xml></xml>’); 02: #SET(s, ‘abc’) In line 2, s is not declared. 2204 Can not find #end for XXX #END is missing for a # command. In ECL template language, some commands need a matching #END: • #FOR • #IF • #IF #ELSE • #LOOP Solution: Add a matching #END. Example: 01: LOADXML(‘<xml></xml>’); 02: #FOR(i) #END is needed for #FOR in line 2. 2205 #ELSE does not match a #IF #ELSE is not allowed outside a #IF structure. Note that #END will end a #IF statement, and #ELSE can no longer be used after the #IF structure is terminated. Solution: Ensure the #ELSE is inside your #IF. Example: 01: LOADXML(‘<xml>dummy</xml>’); 02: 03: #ELSE 04: #DECLARE(s) 05: #END In line 3, #ELSE doesn’t match any #IF statement. 2206 #END does not match a # command A #END is used only to match a # command. This error normally happens if an extra #END is used. Solution: Make sure you use the proper number of #END. Example: 01: LOADXML(‘<xml>dummy</xml>’); 02: 03: #IF(true) 04: #DECLARE(s) 05: #END 06: #END The last #END (in line 6) is extra. 2207 LoadXML failed LoadXML() failed. The reason it failed can be: • The xml file doesn’t exist • The direct xml string is not well-formed. • The xml in the xml file is not well-formed. LoadXML() function can use a direct XML string, e.g., LoadXML(‘<xml>dummy</xml>’); Or it can load from a file, e.g., LoadXML(‘myxml.xml’); Solution: If it is a file, make sure the file exists, and you have the proper permission. Also make sure the xml is well formed. Example: 01: LOADXML(‘non-exist-xml.xml’); If non-exist-xml.xml does not exist, such an error will occur. 2208 #BREAK is only allowed with a #FOR or #LOOP In ECL template language, #BREAK can only be used inside a #FOR or #LOOP. Solution: Make sure it is the right context before you use #BREAK. Example: 01: LOADXML(‘<xml>dummy</xml>’); 02: 03: #IF (true) 04: #BREAK 05: #END #BREAK cannot be used without #FOR or #LOOP in line 4. 2209 #LOOP does not have a #BREAK: an infinite loop will occur In ECL template language, #LOOP is like a while(true) in C/C++. The only way to break out the loop is to use #BREAK. Therefore, a #LOOP without a #BREAK is an infinite loop, and is not allowed. Note: The compiler uses a simple method to detect infinite loops. Don’t depend on it to find infinite loops for you. For example: 01: #LOOP 02: #IF(false) 03: #BREAK 04: #END 05: #END This example has a #BREAK inside the loop, however it is an infinite loop. The compiler is not smart enough to figure out this. However, another simple scheme is used to detect this. See Error C2210. Solution: Use #BREAK to break out of the loop. Normally, #IF is needed to break out of the loop conditionally. Example: 01: LOADXML(‘<xml>dummy</xml>’); 02: 03: #LOOP 04: #END Line 3 and 4 forms an infinite loop. Although it does nothing, it will stuck the compiler if it fails to detect this. 2210 Loops more than max times This error detects more complicated infinite loops other than the cases in error C2209. For each #FOR or #LOOP, the compiler regards it as an infinite loop if it runs more than the predefined max times. The max times is quite large, and should be enough for normal situations. Solution: Make sure it is not an infinite loop. If you are pretty sure you get this error and think it is not an infinite loop (which may never happen to you, just believe me), contact Seisint. We can increase the constant or disable the check. Example: 01: LOADXML(‘<xml></xml>’); 02: 03: #LOOP 04: #IF (false) 05: #BREAK 06: #END 07: #END This example is trick code that is actually an infinite loop. 2211 Unknown # command 2212 Can not fold non pure function 2213 Can not fold non c function 2214 #ELIF does not match a #IF 2216 Bad format for access token 2220 No load function is defined 2221 No store function is defined 2222 No physical length function is defined 2223 Inconsistent logical type 2224 Inconsistent physical type 2225 Store/Load/physicalLength(?) is not defined as func 2226 physicalLength must have physical type 2230 Function is already defined 2231 Attribute is already defined 2232 Invalid library entry 2233 Invalid entrypoint: must be valid C identifier. 2234 Function in service can not specify EXPORT or SHARED 2235 Entrypoint is not defined, default to XXX 2236 Invalid include entry 2237 Serive does need a type 2238 Conflicted attributes defined 2239 Invalid initfunction: must be valid C identifier 2240 Load library failed 2241 Load procedure in library failed 2242 library is not defined (it is an error in template) 2243 Exception occurs when executing service function 2244 Service can not have any parameter 2250 LOADXML is not an attribute: it can only used alone for test purpose 2251 Invalid transform for ITERATE 2252 Invalid transform for JOIN 2255 expected a DATASET(...) 2256 Only one arg supplied in record for INDEX 2257 Parameter to STORED() should be a valid id 2260 A field is mapped more than once 2261 Dataset has no field as required or mapped 2262 A map is not used 2263 Mapping fields type mismatch 2280 token used inside a token definition 2281 Can't refer to a rule inside a pattern 2282 Bad syntax inside a PATTERN() definition 2283 expected pattern 2284 Invalid format for an assertion subpattern 2285 This expression can't be included in a pattern 2286 Can't reference a pattern function in a MATCHED() arg 2287 Self used outside a rule definition 2288 Can only use use() inside a rule definition 2289 :stored etc. used in a pattern attribute 2290 2291 Could not deduce feature from the guard condition 2292 Expected a list of characters 2293 Expected a repeat... 2294 2295 2296 2297 2299 2300 Multi dimension array indexing not supported. 2301 Could not find distributed fields 2302 Only LOCAL merge is currently supported 2303 Field not found 2304 Deprecated 2305 KEYED index is invalid 2306 2307 2308 2309 2310 2311 No longer supported language feature 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2331 2332 2333 2334 2335 2336 2337 2338 2339 Feature not yet enabled on this platform 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 10000 3000 Internal exception Most often this error indicates an internal error of the compiler (a bug), but not always. Some- times, the message correctly reports a problem. In an ideal world, this error should never hap- pen. Solution: Try to work around the situation the message indicates. Contact Seisint when no work- around is apparent. 3001 Too many errors Normally this means either the code is too complicated and has many parsing errors, or the code’s logic confused the compiler, causing cascaded errors. Solution: Try to resolve the errors before this one first, and hopefully there will be fewer errors next time. Comment out some code and try to compile only a portion of the whole may be a good idea sometimes. 3002 Expected ... Normally this indicates a general syntax error. The compiler gives out the symbols it expects based on the grammar. This is the compiler’s last resort for reporting meaningful error messages. Solution: Try to understand the context and the ECL grammer. Don’t take the error message too serious. It may be simply the result of some error before this. 3003 XXX is not valid here Reference Errors C2020 through C2037; these are very similar. 3004 Unexpected end of file encountered The ECL source file reaches the end while the parser is still in the middle of something. This is normally the result of some other error. This also happens during the error recovery produre. Solution: Resolve other errors first, if any. Otherwise, look at the code pointed out carefully. Try to identify problems. Is some structure incomplete? Example: 01: ReverseString4 := TYPE 02: MyType := TYPE 03: EXPORT INTEGER Load(INTEGER x) := x+1; 04: EXPORT INTEGER Store(INTEGER x) := x-1; 05: //END 06: //END In the example, user defined types are not allowed to be nested (see error C2011), so the parser is trying to recover from the error in lines 2-4, but couldn’t find the terminating END when it reached the end of the file. 3005 Whole record not valid here 3006 Except not found in the incoming list 3007 PIPE(name) and project not supported 3100 Field removed from dedup could not be found 3101 Module definition contain an illegal cycle/recursive definition %s 3102 Cannot calculate inversion of PROJECT on STEPPED index%s 3103 Dataset not expected in this context 3104 Equality on field[n..*] must match a similar equality 3105 Join only supports a single x[n..*] comparison 3106 INTERNAL: Serializing extract too complex 3107 INTERNAL: Unexpected '%s' at %s(%d) 3108 INTERNAL: Unexpected type '%s' at %s(%d) 3109 Field '%s' in the map was not found in the dataset parameter 3110 Field '%s' has already been mapped 3111 A field named '%s' was not found in the dataset parameter 3112 File %s not found 3113 Browsing child grouped tables not handled at the moment 3114 INTERNAL: Selected field '%s' does not appear in the dataset 3115 Illegal pattern 3116 Unknown character class [:%s:] 3117 Collation symbols not yet supported 3118 Equivalence class symbols not yet supported 3119 Could not connect to any ECL server 3120 Mismatch in major version number (%s v %s) 3121 Virtual field %s not supported in constant table - please provide a value 3122 Inline DATASET field '%s' cannot be initialized with a list of values 3123 No value or default provided for field %s in inline table 3124 Too many initializers (value %s) for inline dataset definition 3125 Initializer for field %s in inline dataset has the wrong type 3126 Could not constant fold the condition on a IFBLOCK for a inline table Unknown_3005 The parser can not recover from previous error(s) 4000 INTERNAL: Cannot generate code for an empty set in this context 4001 Multi dimensioned sets are not yet supported 4002 ALL cannot be passed to an external service 4003 Indexing ALL is undefined 4004 Cannot use ALL in this context 4005 Too many parameters passed to function '%s' 4006 Cannot define column %s with an unknown length (use alien datatype) 4007 BUILDINDEX can only be used to build fixed width indexes 4008 BUILDINDEX can only be used to build indexes on fixed width files 4009 INDEX does not support fields of type %s 4010 INDEX does not match the dataset being joined 4011 Join condition does not contain sufficient information to use key 4012 Not yet implemented: Cannot use a record structure containing a virtual field (%s) with stored or persist 4013 %s has more than one definition 4014 LIMIT clause can only be applied to a dataset 4015 Cannot resource activity %s a cluster with %d nodes 4016 Unsupported option #WORKUNIT ('%s') 4017 INTERNAL: Unsupported virtual attribute '%s' 4018 Illegal pattern '%s..%s' 4020 THOR must be used for sorting or joining datasets with variable width rows 4021 Substring index %d is outside the field range 4022 RANK/RANKED not supported on list %s 4023 Cannot cast a string of unknown length to another character set 4024 Not enough parameters passed to function '%s' 4025 Index is not supported for type %s yet 4026 RANK has no meaning on an empty list 4027 %s can only be used in a record supplied to a PARSE() command 4028 The parameter to MATCHED(%s) is not found in the pattern 4029 Roxie requires constant filenames - expression %s cannot be computed at deployment time 4030 MATCHTEXT found where MATCHUNICODE was expected 4031 MATCHUNICODE found where MATCHTEXT was expected 4032 Only MATCHTEXT and MATCHUNICODE are valid inside a VALIDATE 4033 Records containing child datasets must be output to a file 4034 Definition of USE(%s) was not found 4035 Definition of USE(%s.%s) was not found 4036 Duplicate definition of %s (use #stored to override default value) 4037 ROWDIFF: Types are not compatible for field %s 4038 FETCH not supported on dataset of kind %s 4039 COUNTER is not legal in this context 4040 XMLTEXT is only legal in a PARSE transform 4041 XMLUNICODE is only legal inside a PARSE transform 4042 Condition on DISTRIBUTE must match the key exactly 4043 Row size %u exceeds the maximum specified (%u) 4044 Select expression should have been hoisted 4045 %s() cannot have a parameter inside a VALIDATE 4046 %s is not supported in roxie queries 4047 Input to MERGE does not appear to be sorted 4048 Expression is too complicated to preload 4049 KEYED(%s) couldn't be looked up in a key. 4050 KEYED(%s) follows unfiltered component %s in the key%s 4051 WILD(%s) follows unfiltered component %s in the key%s 4052 %s could not be implemented by the key 4053 Cannot process zero length rows 4054 Set casts aren't supported yet 4055 Cannot call function %s in a non-global context 4056 Sets of items of unknown length are not yet supported! 4057 #STORED (%s) type mismatch (was '%s' replacement '%s') 4058 Cannot count number of elements in ALL 4059 Can only key fixed fields at fixed offsets 4060 Duplicate definition of %s with different type (use #stored to override default value) 4061 Features are not supported by regex - did you mean repeat() instead of {}? 4062 Option %s not yet supported on child datasets 4063 Grouped tables not yet supported in this context 4064 Nested child queries are not supported yet 4066 #STORED (%s) records must match 4067 ATMOST JOIN cannot be evaluated with this join condition%s 4068 ATMOST JOIN cannot be evaluated with this join condition%s 4069 RIGHT for a full keyed join must be a disk based DATASET 4070 ATMOST(%s) failed to match part of the join condition 4071 "LIMIT(%s, KEYED) could not be merged into an index read" 4072 EXTEND is required on all outputs to NAMED(%s) 4073 Duplicate output to NAMED(%s). EXTEND/OVERWRITE required 4074 OUTPUTs to NAMED(%s) have incompatible types 4075 OVERWRITE is required on all outputs to NAMED(%s) 4076 OVERWRITE/EXTEND should be consistent on all outputs to NAMED(%s) 4078 Insufficient access rights to use embedded C++ 4079 Tag contents syntax <> is not supported by SOAPCALL 4081 RIGHT side of a full keyed join must be a disk file 4082 Expression is not constant: %s 4083 "Unimplemented: Cannot access row blob inside a child query, contact tech support" 4084 Field '%s' with dataset type not supported in csv output 4085 Field '%s' with set type not supported in csv output 4086 Cannot determine the maximum size of the expression 4087 Cannot determine size because variable size dataset is not in scope. Try using sizeof(x,max) 4088 Duplicate definition of %s 4089 Duplicate definition of %s with different type 4090 WILD() doesn't reference fields in key %s 4091 Field %s cannot have both KEYED and KEYED,OPT conditions 4092 KEYED(%s,OPT) follows a WILD() field in key %s 4093 COUNT(,KEYED) cannot be used on a child dataset 4094 Filter for COUNT(,KEYED) did not contained KEYED() expressions 4095 KEYED COUNT used on a non-keyable dataset 4096 Attempting to lookup field %s in a dataset which has no active element 4097 Key condition (%s) is too complex, it cannot be done with the key. 4098 Key condition (%s) requires casts on comparison of field '%s' 4099 Key condition (%s) does not have any comparisons against key fields 4100 No explicit maxlength provided for record %s unary - cannot be performed on a string 4102 %s is not supported inside NOTHOR() 4103 Regular expression parsing doesn't support productions - need to use tomita 4104 Unimplemented: Cannot yet access $<n> inside a child query 4105 Ambiguous default production for rule with multiple inputs 4106 Selector %s is used ambiguously at multiple levels 4107 JOIN%s marked as KEYED does not have a key as the second parameter 4108 ALLNODES(dataset) only legal in roxie and in thor child queries 4109 ALLNODES() is not currently supported on grouped datasets 4110 Library %s cannot contain any workflow actions%s 4111 LOOP body too complicated to evaluate in parallel 4112 Operator %s not supported inside DFAs 4113 MATCHUTF8 found where MATCHUNICODE/MATCHTEXT was expected 4114 THISNODE() can only be used inside ALLNODES() 4115 STEPPED(%s) cannot be evaluated in this context 4116 Parameter to RANGE() is too complex - not currently supported 4117 Project of stepping fields is too complex 4118 Stepped field %s must have a fixed size 4119 Stepped field %s must be at a fixed offset 4120 Expression is too complex to STEP (%s) 4121 Cannot smart step on payload fields 4122 STEPPED range condition could only be matched in one direction 4123 Multiple STEPPED comparisons %s >= 4124 STEPPED condition is not compatible with the sort/merge condition (%s expected) 4125 JOIN condition is not compatible with the sort/merge condition 4126 STEPPED field %s is not keyed 4127 STEPPED field %s does not follow the previous stepped field 4128 Merge order must match all the stepped join fields 4129 Cannot OR together conditions on multiple key fields (%s) 4130 ROW compression can only be used on fixed size indexes 4131 UPDATE cannot be used when the inputs names aren't globally constant 4132 THOR currently only supports LEFT OUTER denormalize 4133 THOR doesn't support DENORMALIZE(%s) 4134 SKIP cannot be used here. It is only valid directly with a transform 4135 Cannot read tables/datasets from a csv file 4136 Argument %s to #workunit must be a constant 4137 Argument %s to #debug must be a constant Unknown_01 Queries libraries must be implemented with a parameterised module 4138 Member %s not defined in module passed as library parameter 4139 Member %s was undefined in a module 4140 Thor does not support HAVING on a non-grouped dataset 4141 %s not currently supported as a child action 4142 Definitions in libraries cannot have side effects (%s) 4143 4144 4145 Index read only supports single STEPPED expression 4146 #%s (%s) creates an illegal recursive reference (%s also being processed) 4147 Insufficient access rights to use PIPE 4148 Parameter to __nameof__ must be a dataset or an index 4149 Expected a UPDATE attribute name 4150 %s @ %d:%d 4152 Index in RANK/RANKED is out of range 4153 INTERNAL: Dataset is not active: '%s' 4154 Could not open thor template '%s' at '%s' 4155 Could not create output '%s' 4156 Dataset '%.80s' does not contain expression '%.50s' 4157 Cannot return a result of this type from a workunit 4158 Could not find dataset %s 4159 Could not find dataset %s (no tables in scope) 4160 INTERNAL: Thor nodes should not be nested 4161 INTERNAL: Missing assignment from transform to %s[%p] 4162 "JOIN%s contains no equality conditions - use ,ALL to allow" 4163 A global DEDUP(ALL) or local hash dedup cannot include comparisons in the dedup criteria 4164 "Global dedup,ALL must have a field to partition" 4165 Hash dedup doesn't support %s 4166 SORTED() used by JOINED must be applied to a DATASET 4167 SORTED() used by JOINED must be applied to a THOR dataset 4168 SORT supplied to COSORT needs to be executed at the same time 4169 WAIT not yet supported 4170 Index has fields %s in scope 4171 Global dataset has fields %s in scope 4172 Create resource library %s failed 4173 Records not assignment compatible 4174 Missing end of placeholder 4175 %s can only be globally or inside ONFAIL or other failure processing 4176 Side-effect%s is not currently supported as a context-dependent dependency 4177 Cannot deduce MERGE transform for global AGGREGATE 4178 AGGREGATE does not include grouping field '%s' in the result record 4179 The calculated maximum row size has overflowed 32bits 4180 %s: %s cannot be dependent on the dataset 4181 Cannot determine the minimum size of the expression 4182 %s used outside of a TABLE aggregation 4183 %s resource added after manifest was finalized 4184 SKIP inside a ROW(<transform>) not supported. It is only allowed in a DATASET transform. 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4512 4513 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4601 Rank/Ranked not supported on variable length strings 4604 DISTRIBUTION() only supported at the outer level 4606 Physical table join condition too complicated 4609 Arbitrary repeats not yet supported! 4610 Tomita doesn't yet support unicode 4611 Patterns are too complicated for Tomita to handle at the moment [%s] 4613 Cannot currently assign to DATASET(SIZEOF(x)) fields 4614 Can only normalize simple child datasets - use other form if more complicated 4615 ROXIE doesn't yet support local activities 4616 4617 STEPPED is not currently supported by thor Unknown_02 string[n..*] syntax is only valid in a join condition Unknown_03 Keyed distribute doesn' support join condition of the form field[n..*] 4618 ONCE workflow items cannot be dependent on other workflow items (including ONCE) 4619 Thor currently only supports the local version of COMBINE 4799 4800 INTERNAL: Clear not supported on LOCAL datasets 4801 INTERNAL: Local datasets cannot be created 4802 INTERNAL: Attempt to access dataset outside of Thor 4803 INTERNAL: Not all Transform targets were assigned to [%s] 4804 INTERNAL: Cannot access stored variable %s in this context 4805 INTERNAL: Evaluate table is not in scope 4806 INTERNAL: Internal function %s not found 4807 INTERNAL: OUTPUT() on a scalar not processed correctly 4808 INTERNAL: Expression too complex - cannot create a DFA 4809 INTERNAL: Unsupported inline query 4813 INTERNAL: Could not resolve file position 4814 INTERNAL: Cannot perform type cast from %s to %s 4815 INTERNAL: Could not find context to evaluate match expression 4816 INTERNAL: Blob translation context not found 4817 INTERNAL: ROWS can not be evaluated in this context 4818 INTERNAL: Expected a parent/container context. Likely to be caused by executing something invalid inside a NOTHOR. 4820 INTERNAL: Could not resolve COUNTER for inline count project 4821 INTERNAL: Graph context not found 4822 INTERNAL: Accessing unserialized stored variable %s in slave context 4823 INTERNAL: Library '%s' shouldn't access work unit temporary '%s' 4824 INTERNAL: Library '%s' shouldn't create work unit temporary '%s' 4825 INTERNAL: Attempting to access graph output directly from a child query 4826 INTERNAL: Inconsistent activity localisation (child %d:graph %d) 4827 INTERNAL: No active parent extract - activity has incorrect localisation? 4828 4829 INTERNAL: Inputs to nary operation have inconsistent record structures 4830 INTERNAL: Linked child rows required without legal context available 4833 4834 INTERNAL: Cannot generated serialized compare function 4835 INTERNAL: Attempt to read spill file %s before it is written 4836 INTERNAL: Dependency within a graph incorrectly generated for hThor (%u) 4837 INTERNAL: Unrecognised compound assign %s