HP 3000 Manuals HP 3000 Manuals
LET Specifies arithmetic operations. Syntax LET destination-variable = arithmetic-expression [,ERROR=label[([item-name])]]; The LET verb is primarily used to perform arithmetic operations.
NOTE At one time the LET verb was also used to manipulate arrays through an optional syntax variation that used the LET OFFSET option. However, the current version of Transact supports subscripting of arrays so that use of the LET OFFSET is no longer necessary. Although it is now recommended that you use subscripts to manipulate arrays, the LET OFFSET option is still available and is described later to aid in maintaining older Transact programs.
LET, unlike MOVE, checks that the data types of items being assigned are compatible with the item to which they are assigned. If necessary, LET performs type conversion. Statement Parts destination- An item name that identifies a location in the data variable register, or Transact-defined name of a special purpose register. The result of the operation is placed in this variable. The destination variable may be any of the names listed below. All of the names except item-name are stored in a special area outside the list and data registers. They are, therefore, not affected by SET(STACK) or RESET(STACK). (item-name The computed or assigned value of item-name. [(sub- The item-name identifies a location in the script)]) data register. The item-name can be subscripted if an array item is referenced. (See "Array Subscripting" in Chapter 3.) LINE An integer, defined as I(5,,2) in Transact/V or defined as I(10,,4) in Transact/iX, that contains the computed or assigned value of the line counter for the current line of terminal display or line printer output. OFFSET An integer, defined as I(5,,2) in Transact/V (item-name) or defined as I(10,,4) in Transact/iX, that contains the offset of an item starting at position zero. PAGE An integer, defined as I(5,,2) in Transact/V or defined as I(10,,4) in Transact/iX, that contains the computed or assigned value of the page counter. PLINE An integer, defined as I(5,,2) in Transact/V or defined as I(10,,4) in Transact/iX, that contains the computed or assigned value of the line counter for the current line of line printer output. STATUS An integer, defined as I(5,,2) in Transact/V or defined as I(10,,4) in Transact/iX, that contains the computed or assigned value of the status register. TLINE An integer, defined as I(5,,2) in Transact/V or defined as I(10,,4) in Transact/iX, that contains the computed or assigned value of the line counter for the current line of terminal display output. arithmetic- A single source, or multiple sources connected by expression arithmetic operators in the format: [-]source1 [operator source2]...[operator sourcen] [-] If the expression is preceded by a minus sign, its negative is assigned. ____________________________________________ NOTE If a positive-only integer is set to a negative number, an error occurs. The value of the specified item will be undefined. Since the outcome is undefined, you should not rely on this procedure to zero out values. Instead, use the ERROR= option to branch to a label and negate the desired fields. ____________________________________________ source An item-name[(subscript)] within parentheses, a numeric constant, one of the Transact-defined names listed above under the description of destination-variable, or one of the functions listed below and described later in this verb. The item-name can be subscripted if an array item is being referenced. (See "Array Subscripting" in Chapter 3.) function ASCII LENGTH LN LOG POSITION SQRT VALUE operator + addition - subtraction * multiplication / division giving the quotient // division giving the remainder ** exponentiation ERROR=label An option to cause branching on arithmetic errors. In [([item- addition to branching and resetting the list register, this name])] option causes the status register to be set to a value that identifies the type of error. (See "Error Branching" later in the description of this verb.) label The label to which the program is to branch when an arithmetic error is encountered. item-name The point to which the list register is to be reset before branching to the error label. If you do not specify an item-name (for example, ERROR=label() or ERROR=label), the list register is reset to empty. If you specify an asterisk (for example, ERROR=label(*)), the list register is not changed. The order of precedence for arithmetic operators is: ** exponentiation // division giving remainder / division giving quotient * multiplication - subtraction + addition You can change the order of evaluation by using square brackets. For example, the following two statements may yield different results: LET (A)=(B) + (C)/(D); LET (A)=[(B) + (C)]/(D); [REV BEG] Invalid Values Cause Differing Results If a problem occurs within a LET statement, the results obtained with Transact/V and Transact/iX may differ. Transact/V will try to substitute a zero for the invalid value and continue with the assignment. Transact/iX will leave the existing value intact and not re-assign the target. This occurs under the following situations: * Unsubscripted source array is shorter than unsubscripted target array. * Item is invalid. * Subscript for a source array is invalid. * Argument for a function is invalid. To avoid the arithmetic differences, use the same number of elements for unsubscripted arrays, and do not use X or U data types in an arithmetic expression.[REV END] Functions The following sections describe the functions available within the LET verb, including parameters and examples. These functions can be used whenever an expression can be used. An additional set of parentheses around item parameters is optional. For example, SQRT(Z) and SQRT((Z)) are both acceptable. A function cannot be embedded or nested in another function. In the following example, the compiler will treat LOG as an array item and generate a warning if LOG is not defined. LET (A) = SQRT(LOG(100.0)); The ERROR= option causes the branch to a label to be taken when specific errors occur while processing a function just as specific errors in a LET statement without a function cause such a branch to be taken. ASCII The ASCII function converts the first character of a string to the number for its ASCII code. The result will be a number between 0 and 255 inclusive. This function is only valid for string constants and data items of type X or U. Syntax ASCII({(item-name[(subscript)])}) {"character-string" } Examples LET (CODE) = ASCII("A"); Before After CODE I(5) 0 65 LET (CODE) = ASCII((ARRAY(2))); Before After ARRAY(2) X(4) BCDE BCDE CODE I(5) 123 66 LENGTH The LENGTH function returns the length in characters of a string by returning the integer index of the position of the last non-blank character in the string. Embedded blanks are included in this count, but trailing blanks are not included. Nulls are considered valid characters and are counted. When calculating the length of an X or U item, the maximum length will be the display length. This function is only valid for string constants and data items of type X or U. Syntax LENGTH({(item-name[(subscript)])}) {"character-string" } Examples LET (COUNT) = LENGTH("__APPLE_"); Before After COUNT I(5) 0 7 LET (COUNT) = LENGTH((ARRAY(2))); Before After ARRAY(2) X(7) ABC_DE_ ABC_DE_ COUNT I(5) 0 6 LET (COUNT) = LENGTH("____"); Before After COUNT I(5) 0 0 LN The LN function computes the natural logarithm of a number.
NOTE Previously, an additional set of parentheses was not allowed around an item parameter in this function. This has been changed so that additional parentheses around an item are optional. For example, LN(A) and LN((A)) are both acceptable.
Syntax LN({(item-name[(subscript)])}) {numeric-constant } Examples LET (RESULT) = LN(100.0); Before After RESULT R(6,2,4) 0.00 4.61 LET (RESULT) = LN((ARRAY(2))); Before After ARRAY(2) R(6,2,4) 10.00 10.00 RESULT R(6,2,4) 0.00 2.30 Errors If the value of the parameter is zero or less, an error message is issued to indicate a computational error has occurred. LOG The LOG function computes the common logarithm to the base 10 of a number.
NOTE Previously, an additional set of parentheses was not allowed around an item parameter in this function. This has been changed so that additional parentheses around an item are optional. For example, LOG(A) and LOG((A)) are both acceptable.
Syntax LOG({(item-name[(subscript)])}) {numeric-constant } Examples LET (RESULT) = LOG(100.0); Before After RESULT R(6,2,4) 0.00 2.0 LET (RESULT) = LOG((ARRAY(2))); Before After ARRAY(2) R(6,2,4) 10.00 10.00 RESULT R(6,2,4) 0.00 1.00 Errors If the value of the parameter is zero or less, an error message is issued to indicate a computational error has occurred. POSITION The POSITION function returns the integer index of the position of the first occurrence of the second string in the first string. Trailing blanks in both strings are ignored. Hence, a string only consisting of blanks cannot be found. If no match is found, then 0 is returned. This function is case sensitive (for example, "a" does not match "A"). This function is only valid for string constants and data items of type X or U. The display length will be used when calculating the length of a data it of type X or U. Syntax POSITION({(item-name1[(subscript)])},{(item-name2[(subscript)])}) {"character-string1" } {"character-string2" } Examples LET (INDEX) = POSITION("GOOD_DOG","Z"); Before After INDEX I(5) 99 0 LET (INDEX) = POSITION((STRING1),"D"); Before After STRING1 X(8) BAD_DOG_ BAD_DOG_ INDEX I(5) 99 3
NOTE In the following example note that the trailing blanks in both arguments are ignored.
LET (INDEX) = POSITION((STRING1),(STRING2(4))); Before After STRING1 X(8) BANANA__ BANANA__ STRING2(4) X(4) NA__ NA__ INDEX I(5) 99 3 SQRT The SQRT function computes the square root of a number.
NOTE Previously, an additional set of parentheses was not allowed around an item parameter in this function. This has been changed so that additional parentheses around an item are optional. For example, SQRT(A) and SQRT((A)) are both acceptable.
Syntax SQRT({(item-name[(subscript)])}) {numeric-constant } Examples LET (RESULT) = SQRT(100.0); Before After RESULT R(6,2,4) 0.00 10.00 LET (RESULT) = SQRT((ARRAY(2))); Before After ARRAY(2) I(5) 64 64 RESULT I(5) 0 8 Errors If the value of the parameter is less than zero, an error message is issued to indicate a computational error has occurred. VALUE The VALUE function returns the numerical value of a string containing the character representation of an integer or a floating point number. Leading blanks are ignored. An initial plus or minus sign is allowed. The number is then terminated by one of the following: (1) the first character that would not be valid in the number; (2) the end of the defined length of the item; or (3) a delimiter defined via the SET(DELIMITER) verb. With Native Language Support, Transact validates numeric data using the thousands and decimal indicators of the language in effect. (See Appendix E, "Native Language Support," for more information.) If a number is not represented in the string, then 0 is returned. Scientific notation (type E) is not parsed in the string. When searching through an item, the last character searched depends upon the data type. For an X or U item, the display length is used to get the last character. For an item defined as I, J, Z, P, K, R, or 9, the value function operates in the same way as a LET assignment. Syntax VALUE({(item-name[(subscript)])}) {"character-string" } Examples LET (NUM) = VALUE("-3A"); Before After NUM I(5) 0 -3 LET (NUM) = VALUE("__+43.21ABC"); Before After NUM R(6,2,4) 0.0 43.21 LET (NUM) = VALUE((ARRAY(2))); Before After ARRAY(2) X(4) 42_3 42_3 NUM I(5) 0 42 LET (NUM) = VALUE("__A3A"); Before After NUM I(5) 0 0 LET (NUM) = VALUE(".52Time"); Before After NUM R(6,2,4) 0.0 0.52 LET (NUM) = VALUE(I); Before After NUM I(5) 0 12345 I I(5) 12345 12345 LET (NUM) = VALUE("123-456"); Before After NUM I(5) 0 123 Syntax Options (1) LET (variable)=[-]arithmetic-expression; Choose this option to place a single value or the result of an arithmetic operation into a location in the data register variable or into one of the Transact-defined names allowed for the destination variable. The following are examples of this syntax option: LET (SUBTOTAL)=(SUBTOTAL) + (AMOUNT); <<Add values of AMOUNT and SUBTOTAL>> <<and place result in SUBTOTAL >> LET (PERCENT)=9.8; <<Set value of PERCENT to 9.8 >> LET (INVERSE)=1/(DIVISOR); <<Calculate inverse value >> LET (CNT)=-(CNT); <<Negate value of CNT >> LET (DEDUCTION)=-[(SUBTOTAL)-(BENEFIT)]; <<The result of subtracting >> <<BENEFIT from SUBTOTAL is >> <<negated and placed in DEDUCTION>> LET PAGE=200; <<Set page counter to 200 >> LET LINE=60-(REMAINING-LINES); <<Calculate value of current line >> LET (STAT) = STATUS; <<Set STAT to contents of status >> <<register >> LET STATUS = STATUS+1; <<Increment value of status register>> LET STATUS = 0; <<Clear status register>> << Set UNIT-PRICE, but if an arithmetic error occurs, branch >> << to CALC-ERROR label and reset list register at UNIT-PRICE. >> LET (UNIT-PRICE) = (SUBTOTAL-PRICE)/(QUANTITY),ERROR=CALC-ERROR(UNIT-PRICE);
NOTE The LET verb is primarily used to perform arithmetic operations on numeric items. No error is generated if a character (X or U type) item is used and processing continues for that character type, but the results may not be as expected. (Use MOVE to handle character items.)
When LET is used with character items, be aware that the display length is used to determine the size of the item. If the destination item is defined with a display length equal to or larger than the source, the entire source is placed in the destination. If the destination is defined with a display length smaller than the source, the source value is truncated on the right when placed in the destination. The following example demonstrates how different display lengths affect the result. SYSTEM T6100; DEFINE(ITEM) SMALL X(5,,6): LARGE X(6,,6); LIST SMALL: LARGE; <<LET uses the display length as the size of the item>> MOVE (SMALL) = "12345"; <<Small has "12345 " in storage >> DISPLAY SMALL; <<Small displays "12345" >> LET (SMALL) = -(SMALL); <<Small has "-12345" in storage >> DISPLAY SMALL; <<Small displays "-1234" >> LET (LARGE) = (SMALL); <<Large has "-1234 " in storage >> DISPLAY SMALL: LARGE; <<Both display "-1234" >> EXIT; (2) LET OFFSET(item-name)=[-]arithmetic-expression (item-name) Identifies an ordinary data item or a child item. [-]arithmetic - Is as defined earlier for the LET verb, except that expression in this context the variables may not be subscripted. This option of the LET verb sets the value of OFFSET for a particular item. It allows you to refer to a child item within a parent item by telling Transact the byte location at which the item begins.
NOTE It is strongly recommended that you address array items by using subscripts. This discussion is included for those dealing with older versions of Transact programs written before subscripting of arrays was implemented. In any case, the LET OFFSET and subscripting should not be used together. Doing so may cause the program to update the data registers in areas outside the limits of the item referenced and could lead to unpredictable results. Since this was previously the only way to manipulate arrays, no error will be generated. (See "Array Subscripts" in Chapter 3.)
By changing the value of OFFSET, you can refer to any child item within the parent item. Suppose an array and its child items are defined as follows:Initially, the OFFSET of YEAR within SALES is 0, which actually refers to byte position 1 of SALES. That is, YEAR(1)= SALES(1), and, therefore, YEAR refers to the first 10 bytes of SALES. To refer to other elements of SALES, you must change the OFFSET of YEAR. You can do it as follows (where element-size is expressed in bytes): LET OFFSET(YEAR)=(element-number - 1) * element-size For example, to point to the third element of SALES, which is 10 bytes long, and then move a value to that element, use the following statements: LET OFFSET(YEAR)= 2 * 10; << (3rd element-1) * element size >> MOVE (YEAR)=(VALUE-STRING); To access and display the second and third positions, use the following statements: SYSTEM TEST; DEFINE(ITEM) SALES 3X(10): YEAR X(10)=SALES(1); PROMPT SALES; DISPLAY SALES; DISPLAY YEAR; LET OFFSET(YEAR)= 1 * 10; <<Access 2nd element of SALES (2-1) >> DISPLAY YEAR; LET OFFSET(YEAR)= 2 * 10; <<Access 3rd element of SALES (3-1) >> DISPLAY YEAR; END; Note that the offset is counted from zero. Thus, to access the second position in SALES, you specify an offset of 1; to access the third position of SALES, you specify an offset of 2. It is possible to step through a parent item using the following form of the LET statement: LET OFFSET(child-item)=OFFSET(child-item)+(byte-length-of-child-item) For example, assuming the same array SALES, you can specify the next child item as follows: LET OFFSET(YEAR) = OFFSET(YEAR) + 10 You can also use the OFFSET option of LET to manipulate complex arrays. Consider the complex array of sales figures shown in Figure 8-1. Its compound items are district, year, and month. Each cell, which is a child item, contains a sales figure in integer format. Note that each value in each cell requires four bytes of storage. This SALES matrix requires the following DEFINE(ITEM) statement: DEFINE(ITEM) SALES-ARRAY X(144): DIST 2 X(72) = SALES-ARRAY: YEAR 3 X(24) = DIST: MONTH 12 X(2) = YEAR: SALES I(4,,2) = MONTH; The fifth line of the DEFINE statement above redefines MONTH as SALES to further identify the data being stored.Figure 8-1. Complex Array of Sales Figures. To locate the position of one SALES element within the array, you must use three LET OFFSET statements. To locate the byte position of the second district of the third year of the seventh month, use the following three LET OFFSET statements: LET OFFSET(DIST) = OFFSET(DIST) + 1 * 72; LET OFFSET(YEAR) = OFFSET(YEAR) + 2 * 24; LET OFFSET(MONTH) = OFFSET(MONTH) + 6 * 2; Since OFFSET leaves the pointer at the last position referenced, it is necessary to either reset the pointer before further manipulation or plan the next OFFSET in terms of the current position. The following statements reset all offsets to zero, representing the position SALES(1,1,1). LET OFFSET(DIST) = 0; LET OFFSET(YEAR) = 0; LET OFFSET(MONTH) = 0; When assigning a value to an array, LET assigns each element in the array to that value. If a subscript is specified, then only that element is assigned the value. All other elements remain unchanged. For example, ARRAY-A is defined as 4X(2), and ARRAY-B is defined as 4I(5,,2). MOVE (TEMP-X) = "ND"; LET (ARRAY-A) = (TEMP-X); <<Sets all elements in ARRAY-A >> DISPLAY ARRAY-A; MOVE (TEMP-X) = "YR"; LET (ARRAY-A(2)) = (TEMP-X); <<Sets second elements only in ARRAY-A>> DISPLAY ARRAY-A; LET (ARRAY-B) = 67; <<Sets all elements in ARRAY-B >> DISPLAY ARRAY-B; LET (ARRAY-B(3)) = 78; <<Sets third element in ARRAY-B >> DISPLAY ARRAY-B; Rounding To determine how rounding is done in Transact, it is necessary to understand how Transact performs arithmetic operations. In general, if you want arithmetic results to be rounded instead of truncated to a desired precision, you should ensure that the [REV BEG]operands, including constants, have at least one more digit of precision than the desired result.[REV END] Transact uses one of three different methods to process arithmetic expressions. The three methods are: * double integer arithmetic * long real arithmetic * packed decimal arithmetic The first two methods, double integer and long real arithmetic, are used only if the values meet particular criteria. When these criteria are not met, the third method, packed decimal, is used by default. Since packed decimal arithmetic is slower than the other two methods, it is advisable to use variables that meet the criteria for one of the other two methods whenever possible. The factors that determine the method to be used are: * Whether the expression consists of a single operation or multiple operations. * Data types of the destination variable and the operands. * The number of decimal places defined for the destination variable and the operands. * Storage length of the destination variable and the operands. 32-Bit Integer Arithmetic To qualify for 32-bit integer arithmetic, an expression must meet all of the following conditions: * The expression must consist of only one operation and that operation can only be +, -, =, or unary minus. * The data types of the destination variable and the operands must be either I or J. Numeric constants cannot be used. * The number of decimal places must be identical in target item and both operands. * Storage length of destination variable and all operands must be 16-bits. When 32-bit integer arithmetic is used, the target item and the operands are converted to 32-bit integers before the operation is performed. The final result is rounded to the precision of the destination variable and then converted back to a 16-bit integer. Although the operands are converted to 32-bit integers before computation, the final result for 16-bit integers should lie between -32768 and 32767. The following is an example of 32-bit integer arithmetic: SYSTEM ARIT02; DEFINE(ITEM) I1 I(4,1): I2 I(4,1): I3 I(4,1); LIST I1: I2: I3; LET (I1) = 45.99; << Packed decimal arithmetic >> LET (I2) = 35.99; << Packed decimal arithmetic >> LET (I3) = (I1) + (I2); << Double integer arithmetic >> DISPLAY; EXIT; When the program is run, the values displayed are: I1 = 46.0 I2 = 36.0 I3 = 82.0 64-Bit Real Arithmetic To qualify for the 64-bit (long) real method of operation, the operands must meet all of the following conditions: * The expression must consist of a single operation which must be +, -, *, /, //, =, unary minus, LN, LOG, SQRT, or **. * The destination variable and the operands must all be of data type R or E. Numeric constants can be used; they are converted to the type of the destination variable. * The storage length of all three variables should be 32-bit or 64-bit. For 64-bit real arithmetic, if the destination variable and the operands are not already 64-bit real, they are converted to 64-bit real before the operation is performed. The final result is converted back to the size of the destination variable. The internal value of the final item may carry more precision than its defined decimal count. Hence, for subsequent 64-bit real arithmetic, the internal value carrying more precision will be used. On the other hand, for subsequent packed decimal arithmetic (see the following discussion), the internal value will be rounded according to the defined precision and then will be used. The internal value will be rounded up for DISPLAY statements. For example, SYSTEM LONGRL; DEFINE(ITEM) REAL1 R(8): << No decimal place. >> REAL2 R(8,2): REAL3 R(8,2); LIST REAL1: REAL2: REAL3; LET (REAL1) = 1440 / 900; << 64-Bit Real >> The display value of REAL1 is 2 (rounded). Internally the value is 1.5555553436279 L+00.. In subsequent 64-bit real arithmetic, the internal value of REAL1 1.555555..... will be used. LET (REAL2) = (REAL1) + (REAL1); In subsequent packed decimal arithmetic, REAL1 and REAL2 will be rounded before computation. LET (REAL3) = (REAL1) * (REAL2) / 3.11; DISPLAY REAL1: REAL2: REAL3; The values displayed are as follows: REAL1 = 2 REAL2 = 3.11 REAL3 = 2.00 Packed Decimal Arithmetic If the values in an expression do not meet the criteria for processing by either the 32-bit integer or 64-bit real methods, then the packed decimal method is used. In this approach, an arithmetic expression is processed according to the rules of precedence described earlier. Before computation, the data types of the destination variable and operands are converted to P - - packed P(27,0,14) - - if the operation is +, -, *, /, //, =, or unary minus. For the remaining functions, such as SQRT, LOG, LN, exponentiation, and so on, the operands and destination variable are converted to 64-bit real. If this function is an intermediate operation, the result is converted to data type P and stacked for continuing with the rest of the expression. Any operands of type R or E that carry greater precision due to previous long real arithmetic are rounded according to the precision defined for packed decimal arithmetic. While an expression is being evaluated according to the rules of precedence, each intermediate computational result is computed to the highest precision of the two operands and the destination item. If the precision of the expression is greater than the precision of the destination item, the result is rounded to the precision of the destination item. For example, 3.0/2.0 would produce 1.5 as an intermediate result, which would round to 2 if stored in a receiving item with no decimal places. Unlike COBOL, Transact does not maintain extra precision just for rounding. Thus some division operations may result in a loss of precision. For example, 3/2 produces 1 instead of 1.5 for an intermediate result if the destination variable has no decimal precision. To ensure that precision is not lost, either the receiving item must have the desired precision (at least one decimal place greater than in the arithmetic expression) or all operands in the entire expression must have the desired precision. For applications that require the destination variable to have fewer or zero decimal places, a two-step arithmetic sequence is recommended. The destination variable of the first LET should have an adequate number of decimal places for processing the whole expression and then the second LET statement should contain a simple assignment (=) to an integer item having fewer or zero decimal places. The following example shows this technique. SYSTEM PAKDEC; DEFINE(ITEM) R1 R(6): << No decimal place >> R2 R(11,5): << More decimal places >> I3 I(9,2); << Fewer decimal places >> Example 1 The LET statement below uses the destination variable R1, which has no decimal places. Compare the final results with the next LET statement. LET (R1) = 11590.0000 * [[6353.6100 / 6354] * [1440/900]]; | ____________ | .9999 | ______ | 1 | _____________ | .9999 | ____________________ | 11588.8410 | _________________ | 11589 (ROUNDED) Example 2 The LET statement below uses the destination variable R2, which has five decimal places. LET (R2) = 11590.0000 * [[6353.6100 / 6354] * [1440/900]]; | ____________ | .99993 | ______ | 1.60000 | _______________ | 1.59989 | ____________________ | 18542.72510 | ________________ | 18542.72510 The LET statement below does a simple assignment to the item I3, which has two decimal places. The result is rounded. This is a packed decimal operation, since data types are different. The internal value of I3 does not carry extra precision. LET (I3) = (R2); | ______ | 18542.73 (ROUNDED) [REV BEG] The next two examples show the effect of using either 2 or 5 digits of decimal precision on the first example above.[REV END] [REV BEG] Example 3 Suppose the following was included in the source file: !PRECISION(2) The original LET statement assigning a value to item R1 (Example 1) is now evaluated as shown below. The major difference from the original evaluation is that 1440/900 is treated as 1.60 rather than 1 which affects all subsequent intermediate results. LET (R1) = 11590.0000 * [[6353.6100 / 6354] * [1440/900]]; | ____________ | .9999 | ______ | 1.60 | _____________ | 1.5998 | ____________________ | 18541.6820 | __________________ | 18542 (ROUNDED) [REV END][REV BEG] Example 4 Now replace the !PRECISION statement above with the following statement: !PRECISION(5) The original LET statement assigning a value to item R1 (Example 1) is now evaluated as shown below. This LET statement is evaluated the same as in Example 2, except that here it is rounded for the final assignment. LET (R1) = 11590.0000 * [[6353.6100 / 6354] * [1440/900]]; | ____________ | .99993 | ______ | 1.60000 | _____________ | 1.59989 | ____________________ | 18542.72510 | __________________ | 18543 (ROUNDED) [REV END][REV BEG]
NOTE The result of the LET statement assigning a value to item R2 in Example 2 is not altered by changing the minimum decimal precision to either 2 or 5. Since the destination variable is defined with 5 decimal digits, all intermediate results automatically maintain at least 5 digits of decimal precision.
For more information about !PRECISION, see "Compiler Commands" in Chapter 9.[REV END]
![[FFN8]](../../pic3k/M017291/FFN8.gif)
![[FFN9]](../../pic3k/M017291/FFN9.gif)