chosing float or dec automatically (+other questions now)

Basil · 2008-09-10 22:27

Hi All,

Not sure if this is dooable but thought i'd ask.
In my dataflash chip, I have a series of decimal and floats all stored in longs (well bytes really, but groups of 4 bytes)
I need to 'play them back' over serial to a com terminal.

Is there was way for the prop to tell wether its reading a decimal number or a float from the dataflash?...(This next part is easy)... and then either send it as decimal, or convert floattostring depening on its decision?

Thanks,

Alec

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Post Edited (Basil) : 9/11/2008 12:38:03 AM GMT

Mike Green · 2008-09-10 23:20

The best you can do is to look at the value in the long and make some assumptions about the likelihood that it's an integer vs. a floating point number. Because the floating point format stores the exponent in the high order portion of the long, it would look like a large integer except for the case of a very very small floating point value. For example, if your integers are greater than 23 bits, they're probably floating point.
Here's the format information: en.wikipedia.org/wiki/IEEE_754

Basil · 2008-09-10 23:25

Hi Mike,

Hmmmm. Good ideas. I think the maximum length my integers will be is 12bits.

So if I do the following, should that work.

temp := value from memory
temp := temp & %1111000000000000
if temp <> 0
temp is a float
else
temp is an int

Thanks,

Alec

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Mike Green · 2008-09-10 23:32

Not quite. Your bitmask has to be %01111111_10000000_00000000_00000000
You also have to decide whether zero is an integer or floating point value. It's a special
case of all zero bits.

Basil · 2008-09-10 23:37

Ahhh ok, thanks Mike.

That makes sense except for 1 part.

If I send all 0's over serial, does it matter if i send serial.str(FloattoString(0.0)) or serial.dec(0)?

Wouldn't the end result still be a whole lot of 0's?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Mike Green · 2008-09-11 00:04

It depends on what's listening on the other end. If the values are being converted back to IEEE
floating point if they have a decimal point in them, you'll still end up with zero values since the
IEEE floating point zero value is the same as the integer zero.

Basil · 2008-09-11 00:15

Great,

Thanks mike [noparse]:)[/noparse]

Will try a few things and see how we go.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Basil · 2008-09-11 00:33

Sorry, I have another question semi related to this:

When saving my data to the dataflash i'm currently doing the following:

DF_Send(Circ_Buf_Start_byte, 32)
DF_Send(DF_next_flight, 32)
DF_Send(flighttime, 32)
...
DF_Send(Pyro_Test, 32)
DF_Send(Pyro_Test, 32)
DF_Send(Pyro_Test, 32)

All these variables are longs and are declared in the VAR section in the same order I write it to the DF.

Are the varaibles actually stored in the RAM in the same order I declare them, so I can loop through the data using the memory address of the variable and incrementing it by 4 each iteration of the loop (I assume id increment 4 bytes to go to the next long, rather than 32bits).

Also, I have a question about the EEPROM object (VarBackup/VarRestore)

In either of these methods, I need to specify a start and end address.
With the end address, is it the address of the first byte AFTER the data I want to save that I pass to the method, or the address of the first byte in the last long that I pass?

As an example, using the above data. Assuming I want to store all longs from 'DF_Next_flight' to 'Pyro_Test[noparse]/noparse][b]2[/b', do I call"
VarBackup(DF_Next_Flight, PyroTest[noparse]/noparse][b]2[/b) or VarBackup(DF_Next_Flight, PyroTest[noparse]/noparse][b]3[/b)

Thanks!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Mike Green · 2008-09-11 00:54

Variables of the same size (byte, word, long) are stored in RAM in the order they're declared. Different sizes are grouped together to minimize padding.

I have no idea about your EEPROM object. There are several in the Object Exchange. Usually the documentation (comments in the source) will explain what parameters are used for what purpose.

Basil · 2008-09-11 00:57

Thanks Mike, so as long as I keep them all as longs they will all be sequential [noparse]:)[/noparse]

I think I answered my own questions with the EEPROM.
This line out of the EEPROM object...

repeat addr from startAddr to endAddr - 1

endAddr-1 would suggest I specify the first byte AFTER the data [noparse]:)[/noparse]

Thanks for the help again,

Alec

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

Basil · 2008-09-11 01:12

Hi, one last questions hopefully [noparse]:)[/noparse]

How would I pass the variable at address 'i' to a method which is expecting an actual variable not an address?

Currently doing lots of DF_Send(Variable, 32) calls where Variable is a long containing a result.

I want to do

i := @First_Long
  repeat while i < (@Last_Long + 1)
    DF_Send([b]Variable at i[/b],32)
    i := i + 4

Its the Variable at i part I cant get, Im missing something obvious I know it...

Alec

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
-Alec

My our page

hippy · 2008-09-11 04:32

  i := @First_Long
  repeat while i < (@Last_Long + 1)
    DF_Send( [b]long[noparse][[/noparse] i ][/b],32)
    i := i + 4

chosing float or dec automatically (+other questions now)

Comments