Does anyone run Postgres without PgBouncer?
Almost every notable managed Postgres provider bundles a connection pooler, giving us a strong hint that connection management is a product problem rather than a user problem.
Almost every notable managed Postgres provider bundles a connection pooler, giving us a strong hint that connection management is a product problem rather than a user problem.
River v0.43 removes test-only transitive dependencies from non-test builds, shrinking Go binaries that include River by about 10 kB.
Postgres database templates make pgtestdb’s per-test database isolation remarkably quick, around 100 ms of setup time, and within spitting distance of River’s schema-based approach.
On how LLMs now drive most of this blog’s workflows, replacing what used to be a lot of small, manual edits.
On how a long stretch of open, contiguous time makes you … less (?) productive.
Making Go’s generic context deadline exceeded errors traceable with small helpers that attribute the operation that timed out and indicate how long the timeout was.
Using json_each and json_extract to get bulk inserts in SQLite with sqlc.
It’s not possible to stop the torrent of LLM-generated content, but how about this as a simple ask: when generating content, keep it short.
Gigabit fiber for less money using this one weird trick: move to Austin.
In 1980, Michael Crichton characters in Congo spoke like cavemen to save satellite bandwidth. It was absurd. Ridiculous! Forty-five years later, we’re doing the same thing with LLMs to save tokens.
Short review of Sofia Coppola’s 2010 movie Somewhere . Unless I miss my mark, it’s the same movie as Lost in Translation ?
I don’t like this app.
Using a coalescable parameter to stub time as necessary in tests, but otherwise use the shared database clock across all operations.
Using built-in net/http facilities to make sure that canceled requests are abandoned immediately to save time and resources.
Fun times with Dropbox’s new ~/Library/CloudStorage location and File Provider API integration.
Writing a wrapper script around MozJPEG to achieve ~80% compression on large JPEGs with little downside.
A safe, succinct test data fixtures pattern using sqlc and validator.
Using Go’s runtime.MemStats and canonical log lines to isolate huge memory allocations to a specific endpoint.
Taking five minutes to write a benchmark so I know which of these I should be reaching for first.
How Postgres’ v7 UUIDs are made monotonic, and why that’s a great feature.