Home/ Questions/Q 8472857
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-10T17:11:28+00:00

In many flavors of SQL, there are three ways you can implicitly set a

  • 0

In many flavors of SQL, there are three ways you can implicitly set a column to NULL on every row insertion. These are:

columnname type NULL
columnname type DEFAULT NULL
columnname type NULL DEFAULT NULL

I.e. the first one sets the NULL flag (as opposed to NOT NULL), second one leaves the NULL flag to the default value, and the third one sets the NULL flag and sets the implicit value as NULL.

I’ve heard about how in Microsoft SQL, the second variation is not exactly the same since you can also customize the default value of the NULL flag for tables or schemas.

But for MySQL, I don’t believe there is such a feature. In addition, in MySQL, columns with the NULL flag, unlike NOT NULL columns, always are implicitly set to NULL on insert if there is no explicit value, even if strict mode is turned on. So that leaves all these column declarations identical.

In my MySQL script, for every NOT NULL column, I specify a DEFAULT value for columns I don’t expect to set in some insertions in my application, to avoid any issues if I were to enable strict mode. Thus, I feel it would be more symmetrical if I were to choose the third variation, even though it is the most verbose and explicit. Are declarations like the third variation common, or does the first or second variation occur more frequently in others’ scripts?

I was considering leaning to the second approach because that’s the alias that MySQL dump uses, but I’m not sure that’s such a good idea since it also dumps integer literals as strings (single quotes) in default clauses of column declarations.

This question may appear to be more opinion than one that has a solution, but I also want to be aware of any potential gotchas that I am not aware of. I may choose to migrate from MySQL to PostgreSQL in the future, and I also could use some advice from those experts, e.g. if PostgreSQL distinguishes these cases like MS-SQL does. Correct me if I am wrong if I made any incorrect assumptions.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Every database that I’ve encountered treats NULLs the way that you describe them. When a column accepts NULL values, then not providing a value on INSERT defaults the value to NULL. I believe this is part of the ANSI standard behavior.

    As for specifying “NULL” itself. That is a matter of preference. The standard does say that a column allows NULLs unless NOT NULL is specified (at the level of the data definition language). So, the NULL itself is unnecesary, and you have a fourth, equivalent option:

    columnname type

    NULLs are quite embedded in the SQL language through the ANSI standard. Your third option is the most explicit, but it also looks a bit unusual.

    If you are planning on being super-consistent throughout your system, then I would take the path you are on. For every column in every table have NULL or NOT NULL. For all columns that take default values, have the DEFAULT statement.

    However, don’t expect other people you work with (now or in the future) to readily follow this example. Many would prefer the fourth option in this case simply because it requires less typing and is how all (or almost all) SQL dialects behave.

    • 0