Re: Lookup tables
Adrian Klaver <adrian.klaver@aklaver.com>
From: Adrian Klaver <adrian.klaver@aklaver.com>
To: Ron Johnson <ronljohnsonjr@gmail.com>,
pgsql-general <pgsql-general@postgresql.org>
Date: 2025-02-04T16:06:00Z
Lists: pgsql-general
On 2/4/25 07:19, Ron Johnson wrote: > How big is the database? A tiny 500MB db just for you can get by with > poor design. (But then, why are you using PG instead of SQLite?) For the reasons listed here: https://sqlite.org/quirks.html In particular: [...] 2. SQLite Is Embedded, Not Client-Server 3. Flexible Typing SQLite is flexible with regard to datatypes. Datatypes are advisory rather than mandatory. 4. Foreign Key Enforcement Is Off By Default [...] > > More importantly, will you ever update the descriptions? Of course not! > Famous last words. Having a separate PK means that you update one row > in one column, while what you've done means that tens/hundreds of > thousands of rows in possibly dozens of tables need to be updated. > > It also means that *you* can easily change things in your ad hoc > database *without forgetting* to update a table. > > This is called an "update anomaly" in relational design theory. > > -- > Death to <Redacted>, and butter sauce. > Don't boil me, I'm still alive. > <Redacted> lobster! -- Adrian Klaver adrian.klaver@aklaver.com