Specific fs-lite path corner-cases/tests

[jamoore]

• ./red.dv and then later ./red.dv results in a directory like: /.-1/a.png
  • I.e. if we move to incrementing part of the server-side path, then this issue should go away.
• cd /tmp; omero import ./a.png; goes under /private/tmp/a.png in ManagedRepo.
  • What should the default be?
    • The value passed in by the user? (possibly)
    • The equivalent of no -p argument to patch (i.e. a.png) (unlikely)
    • The absolute path (also not optimal)
  • Do we try to detect tmp directories and similar? (probably not)
• If the client passes in non-normalized versions of the same path, will these be considered to be in the same directory?
  • Server-side after all paths have been converted to unix-style, they should also be normalized. If this results in a duplicate path, what happens?
    • an exception? (possibly optimal)
    • ignoring one file? (this might not work if we're working via sized arrays)
• What happens if the server-side template is empty? Do we default to the absolute path? How do we communicate that to clients?

[mtbcarroll]

Given answers to the described questions, feel free to have me help code actual test cases.

[jburel]

Referencing ticket #9821 has changed sprint.

[jmoore]

Discussions on the path handling in Dundee last week should have resolved all of these issues. Implementation to happen in other tickets.