· 4 min · PostgreSQL

Locale Cooking: Common Scenarios and Suggestions

Pick your PostgreSQL locale wisely: C.UTF-8 for speed, ICU for multilingual accuracy, or POSIX if you're ASCII-only and desperate.

0 comments

· 2 min · PostgreSQL

Talk Fast: The Speed of Various Collations

Comparing PostgreSQL collation methods reveals surprising performance gaps: ICU beats libc by 5x for language-specific sorts, while the built-in C.UTF8…

0 comments

· 4 min · PostgreSQL

The Doom That Came To PostgreSQL: When Collations Change

When locale libraries change beneath your database, indexes silently break. Learn why this happens and how PostgreSQL now warns you.

1 comment

· 4 min · PostgreSQL

Farm to TABLE: Local(e) Providers

PostgreSQL offers three locale providers—libc, ICU, and the new builtin provider—each with different trade-offs for character collation and localization.

0 comments

· 4 min · PostgreSQL

Speaking in Tongues: PostgreSQL and Character Encodings

Use UTF-8 for your PostgreSQL database. We explain why it's the right choice and when the performance argument for C encoding actually fails you.

1 comment

· 8 min · PostgreSQL

“Gentlemen, this is a 🏈”: Glyphs, Encodings, Collations, and Locales

Characters, encodings, and collations: the fundamentals of how PostgreSQL handles human language.

1 comment

· 1 min · PostgreSQL

Checking Your Privileges, 2

Check out the slides from my latest talk, adapted from the previous blog post on PostgreSQL performance tuning.

0 comments

· 3 min · PostgreSQL

Checking Your Privileges

PostgreSQL grants EXECUTE on new functions to PUBLIC by default, which might let unwanted users run your "secret" functions. Here's how to lock them down.

0 comments

· 1 min · PostgreSQL

“Look It Up: Real-Life Database Indexing” at PgConf.NYC

Slides from "Look It Up: Real-Life Database Indexing" are now live. Learn practical indexing strategies that actually work in production PostgreSQL systems.

1 comment

· 1 min · PostgreSQL

Don’t use ChatGPT to solve problems.

ChatGPT gave wrong answers to 23 of 40 PostgreSQL questions—including one that would corrupt your database. LLMs aren't a substitute for real expertise.

4 comments