Populate sequence containing NaN

Question

Nim language question here. I want to read a series of floats from stdin (this example: 7, 1, 4, 4, nan, 4) and store it in a seq[float] type. The input may contain NaNs. But I fail to integrate such outliers.

My code:

var
  line: TaintedString
  timeSeries: seq[float]

while readline(stdin, line) != false:
  echo timeSeries
  timeSeries.add(parseFloat(line))

The output:

@[]
@[7.0]
@[7.0, 1.0]
@[7.0, 1.0, 4.0]
@[7.0, 1.0, 4.0, 4.0]
@[7.0, 1.0, 4.0, 4.0, nan]
@[nan, nan, nan, nan, nan, nan]

Facing the first NaN, Nim renders all inputs as NaNs. But I want this (last line of output):

@[7.0, 1.0, 4.0, 4.0, nan, 4.0]

How do I solve it correctly in Nim? Documentation says NaNs are supported…

What platform are you on and what back-end compiler are you using? — Anthon, Mar 23 '19 at 06:39
I am using nim 0.19.4 on Linux/i386, backend compiler is gcc and I have the same effect with -d:release switch. Sorry, I omitted the import statement in above snippet. — smartmic, Mar 23 '19 at 08:32

Anthon · Answer 1 · 2019-04-06T07:55:05.600

Since you echo timeSeries before you add the next number, the input of the last line with 4 causes the @[7.0, 1.0, 4.0, 4.0, nan] and it is guesswork what you did after that to get the final output line. Although I doubt there is a valid reason for anything to set every value in the sequence to NaN, it might be that what your input triggered a bug.

I have not been able to reproduce your output with your code (adding the required import strutils) when entering your sequence followed by another 4, nan or empty line (the latter erroring on invalid float).

For easier testing, I put your input in a file input.txt:

and ran the following on the latest stable nim (Nim Compiler Version 0.19.4 [Linux: amd64]) as the latest devel nim (Nim Compiler Version 0.19.9 [Linux: amd64]):

import strutils

var
  line: TaintedString
  timeSeries: seq[float]

echo timeSeries
for line in "input.txt".lines:
  timeSeries.add(parseFloat(line.strip))
  echo timeSeries

(the .strip is only there to handle trailing spaces in the input that were a result of cut-and-paste and sloppy editing)

Both compilers output:

@[]
@[7.0]
@[7.0, 1.0]
@[7.0, 1.0, 4.0]
@[7.0, 1.0, 4.0, 4.0]
@[7.0, 1.0, 4.0, 4.0, nan]
@[7.0, 1.0, 4.0, 4.0, nan, 4.0]

compiling with -d:release did not cause any errors either.

I could reproduce your example, it works for me as well. I will investigate why it failed in my initial try (could not reproduce this quickly though) — smartmic, Mar 23 '19 at 08:49

Populate sequence containing NaN

1 Answers1