Add first version of UNCPath

[jannefiluren]

Work in progress

[codecov-io]

Codecov Report

Merging #12 into master will decrease coverage by 5.51%.
The diff coverage is 2.94%.

@@            Coverage Diff             @@
##           master      #12      +/-   ##
==========================================
- Coverage   76.37%   70.86%   -5.52%     
==========================================
  Files           8        9       +1     
  Lines         419      453      +34     
==========================================
+ Hits          320      321       +1     
- Misses         99      132      +33

Impacted Files	Coverage Δ
src/FilePathsBase.jl	40% <ø> (ø)	⬆️
src/uncpath.jl	2.94% <2.94%> (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 d57a98b...520a96b. Read the comment docs.

[rofinn]

I've commented on julia-vscode/julia-vscode#490 to get a clearer understanding of the specific requirements. Having a better picture of your requirements and what level of UNC support needed should help us:

1. Narrow down the scope of this PR
2. Determine what we should be testing for this path type
3. Determine where the code should eventually live

[davidanthoff]

So I think we should not introduce a new type here, but instead just represent them as WindowsPath, which is what python does. In my mind we "just" need to add logic here to properly parse them, and then make sure all the other functions work properly with them.

[jannefiluren]

I have no preferences how to solve this problem. What I see is that UNC paths seems to be handled poorly with Julia base too. Here is one example:

julia> cd("//hdata/fou/jmg")

julia> pwd()
"\\\\hdata\\fou\\jmg"

julia> splitdrive(pwd())    # Seems wrong to me
("\\\\hdata\\fou", "\\jmg")

Should I open an issue at Julia base for this stuff, or is it not supposed to work?

Furthermore, when writing this PR, I had this issue:

julia> UNC_PATH_START = "\\\\"
"\\\\"

julia> joinpath(UNC_PATH_START, "hdata", "fou")   # Seems correct
"\\\\hdata\\fou"

julia> joinpath(UNC_PATH_START, "hdata", "fou", "jmg")   # Not correct
"\\\\hdata\\foujmg"

The \\\\ is not a part of the path so perhaps that should not work either.

I "solved" this issue in the PR using this line of code instead:

julia> folder = UNC_PATH_START * joinpath("hdata", "fou", "jmg")
"\\\\hdata\\fou\\jmg"

julia> isdir(folder)
true

But that seems more like a hack than a good solution. How would you like me to proceed with this PR? Happy for any advice.

[jannefiluren]

@rofinn and @davidanthoff:

Just some info: I just read this thread on discourse. It might be relevant for handling windows paths better. Look in particular at this pull request. Also look at my comment above.

[davidanthoff]

I think I would just try to copy the semantics from https://github.com/python/cpython/blob/master/Lib/pathlib.py over to here...

[abx78]

Does this work? There are a few packages that cannot be used in Windows corporate environments because depending on this