Minor bug in float to string conversion

When I was using the web interface to lower the nozzle,
the reported nozzle height suddenly jumped up and down.

The problem:

A float with a zero on the right of the comma dot is converted incorrectly,
when the Z height is 0.09 it is presented as 0.9 in the web interface.

The problem location:

The "convert float to string" function in RepRapFirmware.cpp at line 293:

// Float to a string.

static long precision[] = {0,10,100,1000,10000,100000,1000000,10000000,100000000};
char scratchString[STRING_LENGTH];

char* ftoa(char *a, float f, int prec)
{
  if(a == NULL)
    a = scratchString;
  char *ret = a;
  long whole = (long)f;
  if(!whole && f < 0.0)
  {
	  a[0] = '-';
	  a++;
  }
  snprintf(a, STRING_LENGTH, "%d", whole);
  while (*a != '\0') a++;
  *a++ = '.';
  long decimal = abs((long)((f - (float)whole) * precision[prec]));
  snprintf(a, STRING_LENGTH, "%d", decimal);
  return ret;
}

My translation:

The float value is split in two parts, left of the comma dot and right of the comma dot.
When both parts are combined with the desired precision, the %d kills zeroes.

I have tried to fix this and my version of the code looks like this:

// Float to a string.

static const char *formatting[] = {"%d", "%d", "%02d","%03d","%04d","%05d","%06d","%07d","%08d"};
static long precision[] = {0,10,100,1000,10000,100000,1000000,10000000,100000000};
char scratchString[STRING_LENGTH];

char* ftoa(char *a, float f, int prec)
{
  if(a == NULL)
    a = scratchString;
  char *ret = a;
  long whole = (long)f;
  if(!whole && f < 0.0)
  {
	  a[0] = '-';
	  a++;
  }
  snprintf(a, STRING_LENGTH, "%d", whole);
  while (*a != '\0') a++;
  *a++ = '.';
  long decimal = abs((long)((f - (float)whole) * precision[prec]));
  snprintf(a, STRING_LENGTH, formatting[prec], decimal);
  return ret;
}

The bug is a bit hard to trigger on the printer,
I have attached a source file to test it on a PC.

$ ./ftoa
Old ftoa with input 0.000 and precision 2: 0.0
Old ftoa with input 0.090 and precision 2: 0.9
Old ftoa with input 0.123 and precision 2: 0.12
New ftoa with input 0.000 and precision 2: 0.00
New ftoa with input 0.090 and precision 2: 0.09
New ftoa with input 0.123 and precision 2: 0.12

There are warnings about the %d in combination with 'long int',
the %d should probably be replaced by %ld for long int values.

$ gcc ftoa.c -o ftoa
ftoa.c:23:3: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has type ‘long int’ [-Wformat]
ftoa.c:27:3: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has type ‘long int’ [-Wformat]
ftoa.c:44:3: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has type ‘long int’ [-Wformat]

Edit: Sorry for the confusing use of the word comma instead of dot, that was probably nl_NL locale.

Edited 4 time(s). Last edit at 01/12/2014 06:11PM by 3D-ES.

---

RS-Online Ormerod #263, Kossel mini with Minitronics, Prusa i3 MK2

Is this a "localisation" problem or does it hold for UK/us convention of using a dot to separate the decimal?

You might try something like :

snprintf(a, STRING_LENGTH, "%0*d", prec, decimal)

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Quote
bobc
You might try something like :

snprintf(a, STRING_LENGTH, "%0*d", prec, decimal)

Wow, nice!

Thanks

---

RS-Online Ormerod #263, Kossel mini with Minitronics, Prusa i3 MK2

Hi 3D-ES, thanks for that. I had seen the same thing, but assumed it was caused by the bed compensation doing something strange.

An easy way to reproduce the problem is to send a command such as "G92 Z5.09" via the web interface with bed compensation turned off, then the Z value in the interface shows as 5.9.

Are you going to submit this fix to github, or would you like me to do it?

[EDIT: ignore that last bit, from Ian's comment in another thread it sounds like RRP have already accepted your fix.]

Edited 1 time(s). Last edit at 01/13/2014 07:07AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Yup, updated in 0.56 - [github.com]

[github.com] seems to have 0.55, 0.56 and a random dc42 version in it! Hopefully Adrian will spot this and remove the 'extras'...

Ian
RepRapPro tech support