regex lint

The first one is fine, the second one complains.
Aren't these both backed by the same regex lib?
Why the difference?
Which standard should apps in base adhere to?
I prefer the former as allowing empty subexpressions
saves a ton of space on the command line by not
requiring the enumeration of the expansion.

ls | egrep '^.+\.foo(\.bar|)$'

find -sE . -regex '^\./.+\.foo(\.bar|)$' -maxdepth 1
find: -regex: ^\./.+\.foo(\.bar|)$: empty (sub)expression