bug with min and max function

[vrossum]

Hi
I see a weird bug with the min and max functions.
y=min(x,1) is fine
but
y=min(1,x) produces an error.
It seems that any expressions with x need to be in the first argument.

If this is really an error and hard to fix, a note in documentation might help.

As an aside, xmgrace works in both cases, but uses the function 'minof'

[mlund]

Hmm, yes that doesn't sound right. What error do you get? We have generally only fixed issues that allow's xmgr to build on newer systems, but it would be nice to correct that behavior.

[mlund]

I'm not sure if this is the code in question, but here the first argument must be an array of values, while the second argument is an integer defining the number of elements.

xmgr-resurrection/src/setutils.c

Lines 247 to 264 in 9ec1245

	/*
	* compute the mins and maxes of a vector x
	*/
	double vmin(double *x, int n)
	{
	int i;
	double xmin;
	if (n <= 0) {
	return 0.0;
	}
	xmin = x[0];
	for (i = 1; i < n; i++) {
	if (x[i] < xmin) {
	xmin = x[i];
	}
	}
	return xmin;
	}

Inspecting src/pars.c, MIN() does seem to ignore the second argument (1 in your case) and call vmin() which returns the minimum value in the vector. This is in conflict with the manual. Can you confirm this behavior?

[vrossum]

Hi
Thank for taking a look.
y=max(y,2) works as expected (the 2nd arg is not ignored).
max(2,y) throws the error (error in do_compute)

The ideal behaviour would be:

• when a is a vector and b is a single number
  c= max(a,b) should return c_i= max(a_i,b)
• when both vectors
  c_i = max(a_i,b_i)
• and, max should be a symmetric function.

The vmin function, minimum of first n elements seems a bit odd to me.