The clang code avoids an actual memcpy() - I have no objection to extra overhead to account for unknown alignment.
Serial.println(test1.value, 2); //shows: 5.00
Thank you all! I will definitely be implementing using memcpy now that I know that it is the correct way of doing a problem like this. As I am not a compiler developer I am not completely sure of the purpose of memcpy (I looked at the documentation but couldn't completely understand) and would love to hear more about it.
In the end, my main issue was not knowing what my test values should output. The test that I had originally should have printed zero but I assumed that anything other than all zeros should result in a non zero number. I was wrong and should have checked that before posting.
Thank you @6v6gt for catching that operator priority mistake!
Is it "operator priority" or "operator precedence"?
itās a non issue really - everyone gets the point but indeed Itās technically called precedence in the spec
https://en.cppreference.com/w/cpp/language/operator_precedence
Associativity also plays a role in this if you want to dig in.
The word priority is used as well to describe what precedence means - youāll find explanations such as
Descending precedence refers to the priority of the grouping of operators and operands. Considering an expression, an operator which is listed on some row will be grouped prior to any operator that is listed on a row further below it.
Not sure itās an interesting debateā¦
2 Likes
Problem is with the pupils who frequently ask for the reference if the word/terminology/vocabulary is used for the fiist time.
To explain the meaning of precedence, of course, the priority word must be brought in as you have shown in your quote.
In that sense pupils would likely understand priority better than precedence
But itās a good Habit to tell them to look up a word if they donāt understand the meaning, so introducing unusual words is good too.
1 Like
You can use pointer variable for type conversion:
1. Given:
binary32 formatted value:
long int y = 0x42963F36;
What is the corresponding float number?
The following codes could be executed on UNO to get the float number which is: 75.12345886.
float y; //request is placed for four byte wide memory space
long *ptr; //ptr is a pointer; holds beginning address of above space
ptr = (long*)&y; //address goes into ptr. Read/write 4-byte at a time.
*ptr = 0x42963F36; //32-bit data goes into 4-byte wide space requested
Serial.print(y,8);//shows:75.12345886
2. Given
A float number:
float y = 75.12345678;
What is the the corresponding binary32 formatted value?
The following codes could be executed ion UNO to get the value which is: 42963F36.
float y = 75.12345678;
long *ptr;
ptr = (long*)&y;
long m = *ptr;
Serial.println(m,HEX);//shows:42963F36
Thank you for the information.
1. I knew that union structure suffers from type-punning. Now, I know that pointer has also the same problem.
2. Then I don't understand how the following codes that uses memcpy() are valid and superior to pointer/union for type conversion; I have to still remmeber that the array holds lower byte first.
void setup()
{
Serial.begin(9600);
byte intnum[] = {0x36, 0x3F, 0x96, 0x42};
//valid binary32 formatted value of 75.12345678; lower-byte first
float flnum; //4-byte space to hold the above pattern
memcpy(&flnum, intnum, sizeof(intnum)); //array name holds the basse address of the array
Serial.print(flnum, 6); //shows: 75.123458; print() method is doing the magic!
}
void loop()
{
}
3. print() is an overloaded method which recognizes the data type of flnum and accordingly selects the following algorithm/formula to evaluate the Real Value of the expected float number.
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.