Can you please log a jira case to clarify the documentation? (We could consider changing the behavior it is way off the norm.)
> On Aug 17, 2022, at 12:39 AM, Itiel Sadeh <iti...@sqreamtech.com.INVALID> > wrote: > > Sorry, it was a bad wording. > The question was specifically about identifiers that *starts* with a dollar > sign, such as "$jk". > > On Tue, Aug 16, 2022 at 9:53 PM Julian Hyde <jhyde.apa...@gmail.com> wrote: > >> Probably a documentation mistake. Many databases allow ‘$’ and ‘#’ in >> identifiers. For example the following work on Oracle and MySQL: >> >> CREATE TABLE t (i INTEGER); >> SELECT i AS j$k FROM t; >> >> And this works on Oracle: >> >> SELECT i AS j#k FROM t; >> >> Julian >> >> >> >>> On Aug 16, 2022, at 12:11 AM, Itiel Sadeh <iti...@sqreamtech.com.INVALID> >> wrote: >>> >>> Hello all, >>> >>> I've notice the following definition of identifier in parser.jj: >>> < IDENTIFIER: <LETTER> (<LETTER>|<DIGIT>)* > >>> >>> Where the <LETTER> definition includes the dollar sign ($). >>> The Calcite documentation, as well as other databases doesn't permit >>> identifiers to start with dollar sign: >>> >>>> Unquoted identifiers, such as emp, must start with a letter and can only >>>> contain letters, digits, and underscores. They are implicitly converted >> to >>>> upper case. >>>> >>> >>> Is there a reason for this? >>> >>> Thanks, >>> Itiel >> >>