Document path specification for creating new datasets

[adswa]

If I supply a path like /tmp to clone it will refuse to clone a dataset inside, because it attempts a datalad clone <source> /tmp instead of a datalad clone <source> tmp/<dsname>. However, I can't hand-edit the path specification (which I don't think is bad). I'd intuitively expect that when I select /tmp as a path, clone would create /tmp/<ds-i-wanted-to-clone>.

[mih]

Right now, you could choose the path picker and select any path, including a non-existing directory.

[Manukapp]

How would you select a non-existing directory?

Clone was successful when creating a new folder in the path picker, and selecting that new folder (here it is the path written after install(ok)) However, trying to select a new folder does not work, as seen in Screenshot. I also attempted currentdir/non-existingdir (and )

The second screenshot shows the Windows error for non-existing folder

[mih]

Non-existing paths can be "selected" by opening the file-picker (not the directory picker), and then navigate to the parent path of the non-existing target directory. Now you can type the name in the "File name" field and hit return.

Alternatively, one can use the directory picker too, navigate to the parent directory of the target, and then actually create the target directory in the dialog (there is a "plus" button), select that and hit return.

[adswa]

I think we should ensure that this is carefully documented.

[mih]

With #191 we can annotate command to cause a directory picker widget. This would then be used to select an existing directory -- which can also be created directly by the user within the selection dialog.

I think this is intuitive

[Manukapp]

Exploring this issue for create and clone, yet only directory picker is available, no file picker and thus it is impossible to pick a non-existing path. I agree it is simpler in this manner. In addition, Windows file explorer allows to create a New Folder from the directory picker prompt, see screenshot below. I believe we should close this issue (I am not allowed) @mih

[mih]

I agree. Thanks for the feedback!