find and -prune, and GNU and POSIX explanations

Question

My GNU findutils (FSF 2015) says:

-prune
   True; 
   if the file is a directory, do not descend into it.
   If -depth is given, false; no effect.
   Because -delete implies -depth, you cannot usefully use -prune and -delete together.

And Posix:

-prune
    The primary shall always evaluate as true;
    it shall cause find not to descend the current pathname if it is a directory. 
    If the -depth primary is specified, the -prune primary shall have no effect.

I only changed the linebreaks. Two things are remarkable:

Are not both totally ignoring -prune's main effect? i.e. to "collect" the expressions to the left, and leave the rest to a -o - connected expression to the right?

With GNU's man page you at least get directed to an example:

-name pattern
    Base of ...
    ...
    ...   
    To ignore a directory and the files under it, use -prune;
    see an example in the description of -path.
    ...

Well there is a simple example with find . EXPR -prune -o -print. This automatically raises the question why find . prints anything at all.

In retrospect, nicely laid out, it is all there.

Now the related Q:

How are these two -prune explanations (GNU vs. POSIX) related?

To me it looks like posix took a dry technical list with one useful hint and turned it into jargon prose for no reason. It is even misleading.

Because in the end both explanations leave you asking:

Why do I need -o after -prune, if it always returns true?

Here is a tested example: I have some mountpoints to skip, and also want to skip the ramdisk "0" (a zero), and want to list all files containing "test",

find . -xdev -name "0" -prune -o -name '*test*'

Like this it works; it also includes a line ./0, which is actually interesting.

Without the prune-o-trick, I cannot make it work, even with explicit -print.

Answer 1

Are not both totally ignoring -prune's main effect? i.e. to "collect" the expressions to the left, and leave the rest to a -o - connected expression to the right?

No, they’re not ignoring it, because this isn’t -prune’s effect at all, it’s a consequence of find’s operator precedence.

find . EXPR -prune -o -print

is equivalent to

find . \( EXPR -prune \) -o -print

-prune is evaluated if EXPR is true. If EXPR is false, -prune isn’t evaluated, and the compound expression evaluates to false; if EXPR is true, since -prune is always true, the compound expression evaluates to true. The result of all this is that -print is only evaluated if EXPR is false, regardless of what -prune does.

find . -xdev -name "0" -prune -o -name '*test*'

prints ./0 as well as the various *test* matches because of the following rule (quoting the manpage):

If the whole expression contains no actions other than -prune or -print, -print is performed on all files for which the whole expression is true.

On 0, -name "0" -prune evaluates to true, so it is printed.

This rule also explains why find . behaves as it does.

The GNU and POSIX explanations are equivalent:

• “True; ” matches “The primary shall always evaluate as true;”
• “if the file is a directory, do not descend into it.” matches “it shall cause find not to descend the current pathname if it is a directory.”
• “If -depth is given, then -prune has no effect.” matches “If the -depth primary is specified, the -prune primary shall have no effect.”

(quoting the amended find documentation).

-prune doesn’t have to be followed by -o, and there are a few examples which illustrate that in the POSIX documentation for find. This trick in particular is useful to limit find to the immediate contents of the current directory:

find . ! -name . -prune -print