common/funcs/tst.basename.d*: make it compatible with basename(1)

[lwhsu]

Summary

In Solaris, basename(1) and basename(3) both return "." while given an empty
string (""), while in BSD (and Linux) basename(1) returns "" and basename(3)
returns "."

This patch accepts basename() in D script returns "" or "."

While here, also:

• change #!/usr/bin/ksh to #!/usr/bin/env ksh to find ksh in $PATH
• fix quotation inconsistency

Test Plan

sudo kyua test -k /usr/tests/cddl/usr.sbin/dtrace/common/funcs/Kyuafile t_dtrace_contrib:tst_basename_d

[markj]

Thanks.

cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/tst.basename.d

60 ↗

(On Diff #31111)

POSIX says that either output is valid, so both Solaris and FreeBSD are behaving correctly. I would phrase the comment this way: "basename(3) and basename(1) return different results for the empty string on FreeBSD, so we need special handling."

[ngie]

cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/tst.basename.d.out

4 ↗

(On Diff #31111)

While I agree with what you're trying to do below, they're unnecessary stylistic divergences from upstream:

$ echo "\"foo"\"
"foo"
$ echo "\"foo\""
"foo"

Could you please revert all of the style changes in the file?

[lwhsu]

• Update comments
• Revert unnecessary quotation changes

[ngie]

cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/tst.basename.d.out

155 ↗

(On Diff #31140)

"." != "." is always false.

[lwhsu]

cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/tst.basename.d.out

155 ↗

(On Diff #31140)

Yes, and it guarantees the warning in the next line won't be issued. This line is generated by tst.basename.d D script:

printf("if [ \"`basename \"%s\"`\" != \"%s\" -a \".\" != \"%s\" ]; then\n",
    dir[i], basename(dir[i]), basename(dir[i]));

Where we add another acceptable output of basename(1,3) in POSIX.