Nowhere (that I can see) does any documentation "define" that replace returns null on null input to arg 3. Nor is it obvious that any "strict" application of any principle should have it return null.
The documentation says that if arg 2 occurs in arg 1, it is replaced with arg 3. Replacing text with null is problematic, but no problem arises if there is nothing to be done.
Who exactly benefits from a function that fails to return the sensible result that most clearly is available?
Bug reference: 18715 Logged by: Chris Email address: xpusostomos@gmail.com PostgreSQL version: 16.5 Operating system: Linux Description:
OK, but now imagine that at runtime the 3rd argument is null... and we expect it to be null because that variable is not used in that tuple: replace('abcdef', '${m}', null) => null
This is not a bug. Replace is defined to return null on null input (i.e., strict) and that is the behavior you are seeing.
Use Coalesce to convert your null into an empty string.
