Both of these work in Spark SQL:

regexp_replace('1234567890abc', '^(?<one>\\w)(?<two>\\w)(?<three>\\w)', '$1')
regexp_replace('1234567890abc', '^(?<one>\\w)(?<two>\\w)(?<three>\\w)', '${one}')

However, neither work in Databricks SQL. I found that this works for the numeric backreference but I can't find any way to make named backreferences work in Databricks. The problem seems to be around the repl string of regexp_replace.

regexp_replace('1234567890abc', '^(?<one>\\w)(?<two>\\w)(?<three>\\w)', '\$1')

However, when used in the regex itself, it only works without the backslash:

select 'aabc' RLIKE '^(\\w)$1' dollar, 'aabc' RLIKE '^(\\w)\$1' escdollar   -- dollar = true and escdollar = false
select 'aabc' RLIKE '^(?<one>\\w)${one}' name   -- also works

Does anyone know of a way to get named backreferences to work in repl string? Or better yet, to get Spark SQL syntax to work properly in Databricks?