Interpretation of locate -r in various locate implementations

[James Youngman]

I understand that "locate -r" works differently between GNU locate on
the one hand, and both mlocate and slocate on the other.     GNU
locate uses Emacs-dialect regular expressions.   Both mlocate and
slocate use POSIX BREs.

(Although the GNU locate manual page states that "." will match a
newline and that was a deliberate design decision, I notice that the
actual code doesn't do that:

$ printf "hello\nworld\n" > x; ./frcode -0  < x > x.db; for re in
".*hello." ".*hello"; do echo "$re"; ./locate -0  -r "$re" -d ./x.db
| od -c; done; rm -f x x.db
.*hello.
0000000
.*hello
0000000   h   e   l   l   o  \n   w   o   r   l   d  \0
0000014

This is a bug.   The divergence between the code and the manual page
needs to be fixed.  At least one must change.
The Texinfo manual doesn't yet mention newlines in the context of "locate -r".
)

Anyway, on the one hand this divergence of interpretation of regular
expressions is unlikely to be convenient for users, but on the other I
believe it should be possible for tools to correctly handle white
space, even newlines, in file names.    As the community of users of
GNU locate, what are your preferences?

James.