decide whether or not to mask the execute bit. This usually results in
looser-than-necessary execute permissions.
+.SH ALGORITHM
+.IP "I. Argument validation" 0.4i
+.RS
+.IP "a. If any part of the target path contains a symlink" 0.4i
+Return failure
+.IP "b. If there's no default ACL to apply"
+Return success
+.IP "c. If the target is not a (non-hardlink) regular file or directory"
+Return failure
+.RE
+.IP "II. ACL application"
+.RS
+.IP "a. If the target is a directory" 0.4i
+Set the target's default ACL equal to its parent's default ACL
+.IP "b. Set the target's access ACL equal to its parent's default ACL"
+.IP "c. If the target is a directory"
+Return success
+.IP "d. If the target was executable by anyone"
+Return success
+.IP "e. If \fB--no-exec-mask\fR was given"
+Return success
+.IP "f. Unset the user/group/other/mask execute bits"
+.IP "g. Return success"
+.RE
+.P
+The action of apply-default ACL largely mimics what the kernel would
+do if you ran \fImkdir\fR or \fItouch\fR to create a new file. The one
+point of disagreement is on how to mask group-execute permissions for
+files. The kernel will let the \(dqmask\(dq bits prevent group-execute,
+whereas apply-default-acl will explicitly remove the group-execute bits.
+For example,
+
+.nf
+.I $ mkdir herp
+.I $ setfacl --default --modify user:mjo:rw herp
+.I $ touch herp/derp
+.I $ getfacl --omit-header herp/derp
+user::rw-
+user:mjo:rw-
+group::r-x #effective:r--
+mask::rw-
+other::r--
+.fi
+
+In the same situation, apply-default-acl will mask the group \fBx\fR bit:
+
+.nf
+.I $ apply-default-acl herp/derp
+.I $ getfacl --omit-header herp/derp
+user::rw-
+user:mjo:rw-
+group::r--
+mask::rw-
+other::r--
+.fi
+
+The author is of the opinion that the latter is more sensible, if not
+simply more consistent.
+
.SH EXIT CODE
.P
When given a single path, the following codes correspond directly to