I believe that as long as quoted identifiers are allowed, every value allowed as a regular character string should also be allowed as an identifier, including the empty string. (Length limits notwithstanding.)

Some best practices may be against empty string names, but the system should allow it. Also serves as a good "default" name, like the good default character string value would be empty or the good default number value would be zero.

I also believe that database tables with zero columns should be allowed, but that's a separate matter.

-- Darren Duncan

On 2017-10-04 9:51 AM, Simon Slavin wrote:
Given a .csv file which starts like this:

1,RAH,Robert A. Heinlein,"Heinlein, Robert A.",Real Name,
2,IA,Isaac Asimov,"Asimov, Isaac",Real Name,
3,HH,Harry Harrison,"Harrison, Harry",,

Shell tool of this version

SQLite version 3.19.3 2017-06-27 16:48:08

Creates a table with the following columns:

  "ID" TEXT,
  "code" TEXT,
  "name" TEXT,
  "sortOrder" TEXT,
  "" TEXT,
  "AlternativeName" TEXT

I don’t know the externally-governed rules.  I don’t know what rules the 
development team want to follow.  But I’m questioning whether it’s a good idea 
to create a column with that name.  If the dev team think it’s okay, that’s 
fine with me.

I would actually prefer it didn’t include the quotes signs in the column names. 
 They’re not in the .csv file.  But that’s a different matter.

sqlite-users mailing list

Reply via email to