Add file argument to get_nml_value

[jsta]

I am thinking that it would be useful to add a file argument to get_nml_value similar to the other get_* functions.

[lawinslow]

So it would operate directly on a file instead of a loaded nml object?

[jsta]

I think either-or. Consider this block from get_hypsography:

  # if both are passed, glm_nml is used and 'file' is ignored
  if (missing(file) & missing(glm_nml)){stop('glm.nml file path OR glm_nml must be specified')}
  
  if (missing(glm_nml)){
    glm_nml <- read_nml(file)
  }

[lawinslow]

Would it then also overwrite the old original file as well? Basically,

  # if both are passed, glm_nml is used and 'file' is ignored
  if (missing(file) & missing(glm_nml)){stop('glm.nml file path OR glm_nml must be specified')}
  
  if (missing(glm_nml)){
    glm_nml <- read_nml(file)
  }


  #do a bunch of stuff to glm_nml

  write_nml(glm_nml, file)

[jsta]

I don't follow. Why would get_nml_value be doing things to glm_nml? My understanding was that it simply retrieved values for inspection unchanged. Are you thinking of a function that would combine an nml file and and local nml object and write a new nml file (aka "combine_nml")?

[lawinslow]

Ah yeah, ignore that. For some reason I was thinking about a function that muted the file in place. Like we were talking about set_nml.

That being said, wouldn't it make sense to implement the parallel version in set_nml? A file could be specified and it would overwrite the NML with a new one with the new variable set. This would eliminate alot of code that reads, alters, then writes NML files.

[jsta]

I am about to submit a PR with this ability added. I was thinking only of reading values not of writing or setting.

I think what you are talking about is more along the lines of #231?

[lawinslow]

My vote is to parallel this functionality in the set_nml function. But I'll leave this up to @jread-usgs if he wants to merge the PR as-is.

[jordansread]

Awesome work @jsta
I agree w/ Luke that it would be best to parallel this, but in a different PR if possible. I haven't been able to review (in all-day meetings), but will soon.

[jordansread]

Yes, I like this idea, and like the parallel idea @lawinslow proposed for set_nml