errors.digits and/or digits is not honored

[johnblommers]

Consider this console session:

> library(errors)
> options(errors.notation="plus-minus", errors.digits=2, digits=2)
> v1 = 0.555      ; errors(v1) = 0.0555555555
> v2 = 0.44444444 ; errors(v2) = 0     ## Note that we have zero errors on v2
> v3 = 0.333      ; errors(v3) = 0.0333
> v = c(v1, v2, v3)
> v
errors: 0.0555555555 0 0.0333          ## Display not honoring errors.digits=2
[1] 0.56 0.44 0.33                                   ## Correctly honoring digits=2
> data.frame(v)
                 v
1  0.555 +/- 0.056                                  ## Correct
2 0.44444444 +/- 0                              ## value should be printed as 0.44
3  0.333 +/- 0.033                                 ## Correct
> 

What do you think?

[Enchufa2]

The option errors.digits comes into play when the notation (plus-minus or parenthesis) applies, namely, when we are printing a single number or columns of a data frame. That said,

errors: 0.0555555555 0 0.0333          ## Display not honoring errors.digits=2

It shouldn't. However, it should honor digits=2, so this needs a patch.

2 0.44444444 +/- 0               ## value should be printed as 0.44

Agree.

[Enchufa2]

This was easy. Try now, please.

[johnblommers]

The results from the example I posted originally now work as expected ✓

One thing we still see is this sort of thing:

> library(errors)
> options(errors.notation="plus-minus", errors.digits=3, digits=3)
> y1   = 0.71      ; ye = 1e-7 ; errors(y1) = ye
> y1
0.710000000 +/- 0.0000001
> 

I'm not sure but shouldn't the output look like this?:

0.710000000 +/- 0.000000100

[Enchufa2]

The option errors.digits has strict priority over digits in the errors notation, so this output is correct.

Edit: I was looking at your last line. Yes, you are right, let me check. It's really tricky to contemplate all the possible cases.

[Enchufa2]

This is a different issue. I'll open a new one and take a look tomorrow.