ensure lintr runs/installs/tests on R-3.6

[russHyde]

The dev version of xmlparsedata has been updated to fix a parsing bug on R-3.6; before fixing that bug several lintr tests were failing (#384). This change points lintr to the dev version of xmlparsedata

[russHyde]

I'm not quite sure what the problem is here. I suspect the function %||% needs rewriting:

`%||%` <- function(x, y) {                                                        
  if (is.null(x) || is.na(x) || length(x) <= 0) {                                 
    y                                                                             
  } else {                                                                        
    x                                                                             
  }                                                                               
}

===> ?
`%||%` <- function(x, y) {                                                        
  if (is.null(x) || any(is.na(x)) || length(x) <= 0) {                                 
    y                                                                             
  } else {                                                                        
    x                                                                             
  }                                                                               
}

Relates to this change in R3.6 NEWS "Experimentally, setting environment variable R_CHECK_LENGTH_1_LOGIC2 will lead to warnings (or errors if the variable is set to a ‘true’ value) when && or || encounter and use arguments of length more than one." [https://cran.r-project.org/doc/manuals/r-devel/NEWS.html]?

[jimhester]

Maybe this, though we might want to rethink having the is.na() in there at all...

`%||%` <- function(x, y) {                                                        
  if (is.null(x) || length(x) <= 0 || is.na(x[[1L]])) {                                 
    y                                                                             
  } else {                                                                        
    x                                                                             
  }                                                                               
}

[russHyde]

Modified the definition of %||%. This should fix #377

[russHyde]

A similar problem has occurred with the function exclude(). Within the following statement:

file %in% names(exclusions) &&
        exclusions[[file]] == Inf ||
        df$line_number[i] %in% exclusions[[file]]

Presumably exclusions[[file]] can have length 0 or > 1.

Suggest:

file %in% names(exclusions) &&
    length(exclustions[[file]]) == 1L &&
    exclusions[[file]] == Inf ||
    df$line_number[i] %in% exclusions[[file]]

I'm not sure why these errors aren't being detected in my R 3.6.1 environment when I run devtools::test()

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@b5dff71). Click here to learn what that means.
The diff coverage is 100%.

@@           Coverage Diff            @@
##             master    #395   +/-   ##
========================================
  Coverage          ?   82.2%           
========================================
  Files             ?      41           
  Lines             ?    1871           
  Branches          ?       0           
========================================
  Hits              ?    1538           
  Misses            ?     333           
  Partials          ?       0

Impacted Files	Coverage Δ
R/exclude.R	93.75% <100%> (ø)
R/utils.R	71.95% <100%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b5dff71...b4f007f. Read the comment docs.

[russHyde]

This should fix #384 and #391 and #377

Also, #363 related to failing tests in master, so should be fixed now too

[jimhester]

Thanks you can merge this whenever.

[russHyde]

Cheers jim