Interpretation of REPRESENT string format.

[SirWumpus]

I'm looking for clarification concerning the string of significant digits returned in the given buffer to REPRESENT. Typically I'd expect from C printf() family a string formatting 123.45 would typically yield for %e a string 1.23450e02.

But as I read REPRESENT the phrase in bold implies that string returned would be without a decimal point, ie. 123450 with S: 3 FALSE TRUE. Is this a correct interpretation?

From 12.6.1.2143 REPRESENT:

The character string shall consist of the u most significant digits of the significand represented as a decimal fraction with the implied decimal point to the left of the first digit, and the first digit zero only if all digits are zero.

Test cases to illustrate what I understand:

6 SET-PRECISION

t{ 0.1e PAD PRECISION REPRESENT S" 100000" PAD PRECISION COMPARE -> 0 FALSE TRUE 0 }t
t{ 0.4e PAD PRECISION REPRESENT S" 400000" PAD PRECISION COMPARE -> 0 FALSE TRUE 0 }t
t{ 0.6e PAD PRECISION REPRESENT S" 600000" PAD PRECISION COMPARE -> 0 FALSE TRUE 0 }t

t{    0.0   PAD PRECISION REPRESENT S" 000000" PAD PRECISION COMPARE -> 0 FALSE TRUE 0 }t
t{  123.45  PAD PRECISION REPRESENT S" 123450" PAD PRECISION COMPARE -> 3 FALSE TRUE 0 }t
t{ -123.45  PAD PRECISION REPRESENT S" 123450" PAD PRECISION COMPARE -> 3  TRUE TRUE 0 }t
t{  0.12345 PAD PRECISION REPRESENT S" 123450" PAD PRECISION COMPARE -> 0 FALSE TRUE 0 }t
t{ -0.12345 PAD PRECISION REPRESENT S" 123450" PAD PRECISION COMPARE -> 0  TRUE TRUE 0 }t

2.718281828459 FCONSTANT e
3.141592653589 FCONSTANT pi
6.62607015e-34 FCONSTANT planck

t{ e        PAD PRECISION REPRESENT S" 271828" PAD PRECISION COMPARE -> 1  FALSE TRUE 0 }t
t{ pi       PAD PRECISION REPRESENT S" 314159" PAD PRECISION COMPARE -> 1  FALSE TRUE 0 }t
t{ planck   PAD PRECISION REPRESENT S" 662607" PAD PRECISION COMPARE -> -33 FALSE TRUE 0 }t

[ruv]

string returned would be without a decimal point

Yes, the string returned by represent does not contain a decimal point, as well as a minus sign (if applicable).

An interesting question is what exponent value it should return in the following cases:

• the first input parameter on the data stack is zero;
• the input floating point value is zero.

I think, when the first input parameter is zero, the floating-point number should be rounded to the nearest integer rationale that is a power of ten, and the returned parameter n is that power exponent.

[SirWumpus]

I think there is a problem with the wording about caddr u. Every where else the pair caddr u, the u is the length of a string or the size of a buffer. Expressing u in terms in the size of the significand is a little confusing. If u is zero then the buffer caddr should be empty; not sure n has any meaning without at least the first digit of the significand in that case.

In the case of 0.0 should the buffer contain just 0 or padded with zeros eg. 00...0u; how many zero's do you need to make a point (outside of politics).

I think REPRESENT seriously needs test cases to demonstrate what is expected.

[ruv]

If u is zero then the buffer caddr should be empty;

c-addr is an address, it cannot be empty. The buffer itself is empty in the sense that its length is zero.

not sure n has any meaning without at least the first digit of the significand in that case.

In this case n could mean the decimal order of the number. Namely, if rounding to nearest is used, then n (which is a member of the data type n) is the value that minimizes the function f(n) = abs( 10^n - abs(r) ).

---

In the case of 0.0 should the buffer contain just 0 or padded with zeros eg. 00...0u;

According to the spec, the buffer shall be padded with zeros: "The character string shall consist of the u most significant digits".

---

I think REPRESENT seriously needs test cases to demonstrate what is expected.

Yes, sure!