Thank you, this is all really helpful. I don't normally work with 'working directories' that are different from the executable directory (just launching the executable specifically or by shortcut), and I've very little experience with Mac and Linux systems.

Thank you very much!

Rather than start a new thread, here's a closely related question: Is an empty file extension the same as no file extension?

Example: Is file the same as file. ? One has a no period and no extension, the other has a period with an empty extension. Are they treated identically by modern OSes (> Win XP), including Linux and Mac OSX?

I ask this because: Given a path by a string, how does one know if the path ends with a file or a directory?

If the final string segment ends with a '/', you know it's a directory.

If the final string segment contains a period, you are fairly sure it's a file (since folders don't usually, but can, contain periods).

But if the final string segment contains neither a period nor ends with a slash, it's very ambiguous (but assumed to be a file?).

Files and Folders aren't just differentiated by 'extension'. In fact, extensions aren't really anything special - they're just part of the filename (windows can be configured to hide that part of the filename, but in reality, the extension is just a string).

You can try this yourself, on linux, if you type "ls -l" you'll see a list of properties preceeding each file, something like this for executables:

-rwxr-xr-x

or this:

-rw-r--r--

for non-executables. The properties are separated into categories - owner, group, and other - and each is given permission to read (r), write (w) or execute (x) a file.

see chmod man pages if you want more explanation: http://ss64.com/bash/chmod.html

A folder isn't like a file at all. It's an entry in the file system that simply says that there's a folder there. You rename a folder to something.exe and it still won't be executed (same on linux too)

Btw, a folder's properties on linux look like this:

drwxr-xr-x

That first letter (d) indicates its a folder on the filesystem.

Renaming a file that's 'executable' to whatever extension will stilll let you run that file from the command shell in linux. Windows is another story, that I'm not 100% sure of. It seems like it does treat files based on their extensions (for executables I mean).

Oh, and filenames in linux can't contain /. (though they can contain all other sorts of unreadable and impossible to use characters.. including newline..yeah...)

For the most part, given a path as a string you don't know if it's a file or a directory. You need to query the file system. Ex: with the stat() system function on POSIX systems, the is_directory() function in boost::filesystem, etc.

Thanks, that's what I was wanting to know - whether there was a way to be reasonably confident merely from the string without querying the file system itself. Other than ending in a '/' guaranteeing it's not a file, there's not much you can go off of!

It would've been great if OSes enforced a period for filenames, even if the extension was left blank, or maybe used a separator to indicate where the path ends and the file begins. Maybe ':', like: my/path/to/file/:filename

Ofcourse, extensions don't really confirm a file is of any specific format anyway, so maybe that's just band-aiding an imperfect, but functional, system.