gensub disagrees with its sub-gsub equivalent

[Miguel Pineiro Jr.]

Hello Arnold and Co.,

gawk's gensub disagrees with its corresponding sub/gsub. Perhaps
an unintentional application of the regular expression escape
sequence rule which discards backslashes that don't introduce a
valid sequence?

BEGIN {
        s = "aa"
        print gensub(/a/, "\\x", 1, s)
        sub(/a/, "\\x", s)
        print s

        s = "aa"
        print gensub(/a/, "\\x", "g", s)
        gsub(/a/, "\\x", s)
        print s
}

I tested with an unmodified version of the current master branch:
* master 8e8b8cf2 A few more in xhe.po.

gawk output:
xa
\xa
xx
\x\x

BSD branch of onetrueawk and busybox:
\xa
\xa
\x\x
\x\x

I hope everyone is well,
Miguel