I am trying to use the 'Last attribute with a 2D array in Ada, but I can't seem to find the correct syntax to do so.
I know that if I have a 1D array/vector I can use A'last or A'last(n) where n is the nth dimension. But if I do the following
type UnconstrainedArray_2D is array (Integer range <>, Integer range <>) of Integer;
function temp(tempIn : in Integer;
Table : in UnconstrainedArray_2D) return Integer is
tempTable : UnconstrainedArray_2D(0..tempIn, 0..tempIn);
begin
for i in 0..tempTable'last(1) loop
for j in 0..tempTable'last(2) loop
tempTable(i, j) := Table(i,j);
end loop;
end loop;
end temp;
I get the following compile time error:
Storage_Error stack overflow (or erroneous memory access)
So what am I doing wrong?
I am using GNAT Pro 6.4.1 on Linux.
I'd be very surprised if you got a compile-time Storage_Error on that code.
I've grabbed a copy of your code and modified it as follows; it compiles without error using GNAT (gcc-4.4):
procedure Array_2D is
type UnconstrainedArray_2D is array (Integer range <>, Integer range <>) of Integer;
function temp(tempIn : in Integer;
Table : in UnconstrainedArray_2D) return Integer is
tempTable : UnconstrainedArray_2D(0..tempIn, 0..tempIn);
begin
for i in 0..tempTable'last(1) loop
for j in 0..tempTable'last(2) loop
tempTable(i, j) := Table(i,j);
end loop;
end loop;
return 42; -- added this
end temp;
begin
null;
end Array_2D;
(Note that I had to add the missing return statement in temp.)
Your syntax for the 'Last attribute (not "command") is correct, but since Ada arrays can have arbitrary lower and upper bounds, it's better to use the 'Range attribute instead:
for i in tempTable'Range(1) loop
for j in tempTable'Range(2) loop
tempTable(i, j) := Table(i,j);
end loop;
end loop;
As for the Storage_Error exception, that could easily happen at run time (not compile time) if you call your temp function with a very large value for tempIn. Remember that it has to allocate enough space to hold tempIn**2 Integer objects. Presumably you've also created another UnconstrainedArray_2D object to be passed in as the Table parameter.
It's conceivable that the compiler itself could die with a Storage_Error exception, but I don't see anything in your code that might cause that.
Show us a complete (but small) program that demonstrates the problem you're having, along with the exact (copy-and-pasted) error message. Please distinguish clearly between compile-time and run-time errors.
Your tempTable might have a range of 0..tempIn, but you don't know what range your Table has.. They could be of different length, too.
You would have to check that the length is the same and then use relative indices, like this:
function temp(tempIn : in Integer;
Table : in UnconstrainedArray_2D) return Integer is
tempTable : UnconstrainedArray_2D(0..tempIn, 0..tempIn);
begin
if tempTable'Length (1) /= Table'Length (1) or else
tempTable'Length (2) /= Table'Length (2)
then
raise Constraint_Error; -- or something else
end if;
for i in 0 .. tempTable'Length (1) - 1 loop
for j in 0 .. tempTable'Length (2) - 1 loop
tempTable(tempTable'First (1) + i, tempTable'First (2) + j) :=
Table(Table'First (1) + i, Table'First (2) + j);
end loop;
end loop;
end temp;
that way it is ensured that both tables are same length and all indices are valid.
If your tempTable is allowed to be smaller than Table, simply adjust the length check to >. The indices would still be valid.
I don't see an actual value for tempIn set. If the value for tempIn coming into the function temp has not been properly initialized or explicitly set, then the value in tempIn could be anything and probably not something you would like.
I was thinking of a default value. (probably shouldn't post when I am not feeling well :-)
Related
In Ada, how do you loop across any array by any index, starting after the first index? By "any array", we mean empty arrays and arrays whose index is not an integer, too.
EDIT: If an initial conditional is required to handle corner cases - for example: empty arrays - that is fine.
EDIT: Specified "any index" instead of just the second.
Remember that the Ada 'for' loop does not include an increment operator. Instead it iterates through a range of values. The range of values may be the entire set of array indices or it may be a contiguous subset of that range.
The easiest way to accomplish this is to declare an unconstrained array type and then pass the slice of the array you want to process.
procedure main is
type Days is (Mon, Tue, Wed, Thu, Fri, Sat, Sun);
type Day_Counts is array(Days range <>) of Integer;
function Total(Item : in Day_Counts) return Integer is
Sum : Integer := 0;
begin
for Day in Item'Range loop
Sum := Sum + Item(Day);
end loop;
return Sum;
end Total;
Weekly_Counts : DayCounts := (1,2,3,4,5,6,7);
Weekly_Sum : Integer;
Weekend_Sum : Integer;
begin
Weekly_Sum := Total(Weekly_Counts);
Weekend_Sum := Total(Weekly_Counts(Sat..Sun));
end main;
The value placed in Weekly_Sum will be the sum of all 7 array elements. The value placed in Weekend_Sum will be the sum of only the Sat and Sun values.
In Ada you can specify which indexes to iterate over:
-- Declarations
Start : Index_Type;
Finish : Index_Type;
-- Usage
Start := -- Pick your start
Finish := -- Pick your end
for Index in Start .. Finish loop
-- do your stuff
end loop;
-- Example
with Ada.Text_IO; use Ada.Text_IO;
procedure Test is
type Index_Type is (Red, Blue, Green);
type Array_Type is array(Index_Type range <>) of Integer;
My_Array : Array_Type(Index_Type'Range) := (1,2,3);
Start, Finish : Index_Type;
begin
Start := Blue;
Finish := Green;
for Index in Start .. Finish loop
Put_Line(My_Array(Index)'Image);
end loop;
Put_Line("Hello World");
end Test;
where Start and End can be any index type you want. Or you can just iterate over all of them if you want and let the compiler determine what the first and last are.
This works for any type that can be an index of an array (Enumerations, Integers, etc.).
For any index type you can do things like:
Index_Type'First
Index_Type'Last
Index_Type'Succ(value)
Index_Type'Pred(value)
My_Array'Length
My_Array'Range
among many others. These should allow you to do what index math you need (again independent of the index type). See some examples below
for Index in My_Array'Range loop
if Index /= My_Array'First then
-- do stuff here
end if;
end loop;
if My_Array'First /= Index_Type'Last then
for Index in Index_Type'Succ(My_Array'First) .. My_Array'Last loop
-- Do your stuff
end loop;
end if;
I have some codes like follow:
keys: LINKED_LIST[K]
...
test
local
tempK:K
tempI:INTEGER
do
...
across
keys as cursor
loop
tempK := cursor.item
if tempK ~ 1 then
tempI := tempK
end
end
...
end
"cursor.item" is type of "K". However, the real value inside there is integer type.
thus, "if tempK ~ 1 then" works fine. But "tempI := tempK" doesn't work.
How can I convert tempK's K type to integer? so that it can compile?
In this particular case, where you know that tempK is 1, tempI := 1 will do.
If the idea is to initialize tempI as soon the values stored in the list are of type INTEGER, there are several ways. One is to use an object test:
if attached {INTEGER} tempK as i then
tempI := i
end
However, in this case the test is performed for every element, i.e. inefficient. Changing the code to test for the list type before the loop will help:
if attached {LINKED_LIST [INTEGER]} keys as integer_keys then
...
across
integer_keys as cursor
loop
tempI := cursor.item
end
...
end
If the only operation in the loop is the assignment, the equivalent code is to take just the last element of the list:
...
if not keys.is_empty and then attached {LINKED_LIST [INTEGER]} keys as integer_keys then
tempI := integer_keys.last
end
...
Instead of specialization, the code could also be generalized to take a generic agent that will be passed the key, and the client will supply the procedure to handle the key. But this might be too much, depending on what is the purpose of the task you are solving.
I've been trying to make a program that will ask user to input elements of an array and then use that array to make a new one whose elements would be every 2nd element of inputted array. This is what I was writing:
program Keanu;
uses crt;
type Arr=array of integer;
var n,i:integer;
A,C:Arr;
begin
writeln('--Enter desired length of array--');
readln(n);
setlength(A,n);
setlength(C,n);
writeln('Elements of array A:');
for i:=1 to n do
readln(A[i]);
writeln('Elements of array C are:');
i:=1;
while (i<=n) do
begin
c[i]:=a[i];
i:=i+2;
end;
write('C = {');
for i:=1 to n do
begin
if c[i]=0 then continue else
begin
write(c[i],' ');
end;
end;
write('}');
readln;
end.
But as you can notice this is far from efficient way to make this program do the job.
First, because my new array will contain blank/empty elements(zeros) which I simply ignored with continue statement and I dont want to do that if possible.
Second,I have problem when inputting an even number for array length.Last element of new array in output window is very small,negative number and he shouldn't be there at all.I know this has to do something with my counter "i" crossing into "undefined" indexes of array.
I also tried replacing while loop with some variations of:
for i:=0 to n do
c[i]:=a[2*i-1] ;
Which is more elegant way but I still get , besides desired result , those large numbers , again because of crossing limits of an array.I suspect this has to be done with proper steps of how new array is made and moving those elements next to each other with no blank elements.
So, if anyone can give me some solutions of how to get these loop steps and limits into right order and make efficient,shortest algorithm, and preferably without using while loop if possible ,and definitely without ignoring those blank elements of new array.
Declaring variables by one character A, C: array of integer is a bad practice. The name of variable will tell you about it's type and meaning. (highly recommended)
And read about Hungarian notation, then you'll understand why TArr and arrA instead of Arr and A. Not necessary, but recommended. (also recommended)
As for the second arrC, you can use operator div to make it twice smaller than the first array arrA.
You can also crete a new variable that will be n div 2, in order not to change n div 2 in the whole code (good practice):
nDivided := n div 2;
SetLength(arrA,n);
SetLength(arrC, nDivided);
This will make your program quite a bit efficient: you'll save n - (n div 2) * sizeof(integer) bytes.
Here are both cases (for even and odd N). No "blank elements" and no "very small, negative number in the end of new array(-32768)".
N is 6 N is 5
Elements of array A: Elements of array A:
arrA[1]=1 arrA[1]=1
arrA[2]=2 arrA[2]=2
arrA[3]=3 arrA[3]=3
arrA[4]=4 arrA[4]=4
arrA[5]=5 arrA[5]=5
arrA[6]=6
Elements of array C are: Elements of array C are:
arrC[1]=2 arrC[1]=2
arrC[2]=4 arrC[2]=4
arrC[3]=6
Anyway, here is code (a little changed). Also not very good (efficent) but it does exactly what you need.
program WorkingWithArrays;
uses crt;
type
TArr = Array of Integer;
var
i, n: integer;
arrA, arrC: TArr;
begin
writeln('Enter the length of array:');
readln(n);
SetLength(arrA,n);
SetLength(arrC, n div 2);
writeln('Elements of array A:');
for i:=1 to (n) do begin
arrA[i]:=i;
writeln('#> arrA[',i,']=', arrA[i]);
end;
writeln('Elements of array C are:');
i:=1;
while (i <= n div 2) do
begin
arrC[i]:=arrA[i+i];
i:=i+1;
end;
for i:=0 to (n div 2) do begin
if arrC[i]=0 then continue else begin
writeln('#> arrC[',i,']=', arrC[i]);
end;
end;
readln;
end.
// compiled here: http://www.compileonline.com/compile_pascal_online.php
I got the following code which is supposed to correctly acess the matrix positions, however I'm always getting this access violation...
var
tabla : array of array of string;
....
implementation
SetLength(tabla, minterms.Count+1, minterms_essentials.Count+1);
for i := 0 to minterms.Count-1 do
begin
tabla[i+2,1] := minterms[i];
end;
for i := 0 to minterms_essentials.Count-1 do
begin
tabla[1, i+2] := minterms_essentials[i];
end;
end
Basically, I'm generating a table and in the loop I'm trying to fill the column tags and the row tags in second loop. As long as I know, arrays start in 1.
tabla[1][1] would an unoccupied index, that's why I'm ain't touching it.
Why the access violation?
In Delphi, dynamic arrays (any array that you can call SetLength on, as opposed to declaring its bounds at compile time like array[1..5] of integer) are indexed starting from 0, not from 1. So by treating your array as if it used 1-based indexing instead of 0-based indexing, you're overflowing the bounds of the array and trying to write to memory that's not allocated to you, which can cause an access violation.
Dynamic arrays always start at 0.
Because of
SetLength(tabla, minterms.Count+1, minterms_essentials.Count+1);
the highest possible first index of tabla is minterms.Count.
Now, think about
for i := 0 to minterms.Count-1 do
begin
tabla[i+2,1] := minterms[i];
When i is minterms.Count-1, i+2 is minterms.Count+1. Hence, you try to access tabla[minterms.Count+1]. But this doesn't exist, because, as we saw, the maximum possible first index of tabla is minterms.Count.
Hence you try to access something that doesn't exist.
Arrays start at zero and the max. index is Count-1
Therefore, if minterms.Count equals three, setlength(..., 4) ==> Index between 0 and 3.
for i := 0 to minterms.Count-1 do
will do just fine, but change i+2 into i in the loops body.
I need to remove element from array. I have tried to use array.delete(n) function, but it deletes all elements from identifier n. How to just delete exact element n ?
For example if array is 1 2 3 4 5, and n = 3, after delete it should look like following 1 2 4 5.
My code so far :
DECLARE
/* declare type array */
TYPE number_index_by_number IS TABLE OF number INDEX BY binary_integer;
v_n NUMBER := &sv_n;
v_m NUMBER := &sv_m;
v_min Number;
v_tmp Number;
v_array number_index_by_number;
v_sorted_array number_index_by_number;
begin
for i in 1..v_n
loop
v_array(i) := dbms_random.value(1,1000);
end loop;
for j in v_array.first..v_array.last
loop
DBMS_OUTPUT.put_line('v_array('||j||') :'||v_array(j));
end loop;
<<i_loop>> for i in 1..v_m
loop
/*set first array value to variable min*/
v_min := v_array(1);
v_tmp := 1;
<<j_loop>> for j in v_array.first..v_array.last
loop
DBMS_OUTPUT.put_line('v_array('||j||') :'||v_array(j));
if (v_min > v_array(j)) THEN
begin
v_min := v_array(j);
v_tmp := j;
DBMS_OUTPUT.put_line(j);
end;
end if;
end loop;
/*problem is in at this line*/
v_array.delete(v_tmp);
v_sorted_array(i) := v_min;
end loop;
for i in v_sorted_array.first..v_sorted_array.last
loop
DBMS_OUTPUT.put_line('v_sorted_array('||i||') :'||v_sorted_array(i));
end loop;
end;
I cannot reproduce any of the behaviour you describe. I could not get the delete collection method to do anything other than what it is documented to do.
However, there are a few errors in your code that could do with being tidied up.
Firstly, I should point out if you delete an element with key 3 from a PL/SQL associative array, there is then nothing with key 3 in the array. The remaining values don't 'shuffle' down to fill the gap. If there was an element with key 4 before the delete, the same element will still have key 4 afterwards. As a result, if you delete element j from a PL/SQL associative array v_array and then attempt to get v_array(j), you will get a 'no data found' error. You should check to see whether the element exists, using v_array.exists(j), before attempting to get a nonexistent element.
Secondly, the element with index 1 may get deleted before the last iteration of the outer loop. If this happens, v_array(1) will fail with a 'no data found' error. It would be better to assign NULL to v_min and v_tmp at the start of the loop, and assign to them during the loop if v_min is NULL or greater than v_array(j).
Finally, it seems your code returns the v_m smallest numbers from v_n. It would be worth verifying that v_m is less than or equal to v_n, as otherwise this doesn't make sense.
I'm affraid you cannot use a built-in method like this.
Instead of you shoud create a temp array to collect the elements prior to and afterwards the selected one from the original array, and return the temp array.