3 ms·

"Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should." While interesting, the advice offered in this p

"Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should." While interesting, the advice offered in this post is generally bad, or at least in complete. OK, so, you "cut out the middle tier..." what now? Are web clients connecting to PostGRES directly? Will PostGRES handle connection authentication and request authorization? Logging that a particular user looked at particular data? Can the client cache this data? If so, for how long? Even taking the author's premise, this is not a good pattern - a simple JSON get request still comes with a bunch of other stuff that this doesn't bother to address. But the premise is wrong - few applications just spit out JSON like this, they also have to handle updates, and there's business logic involved in that. Data needs to be structured in a way that's reasonable for clients to consume, which isn't always a row of data (maybe it should be collapsing all values in a column, so that in the JSON the client just gets a JSON array of the values).

simonw 5y ago
I worked on a Django project recently where we were generating GeoJSON from data in a PostgreSQL table. We investigated using tricks like this to generate the JSON directly from a SQL query - my notes here: https://til.simonwillison.net/postgresql/constructing-geojson-in-postgresql https://til.simonwillison.net/postgresql/constructing-geojso... The performance was a big enough improvement that, had we been bottlenecked on this particular operation, we would have shipped this (I'm afraid I don't have notes on benchmarks). We only decided not to ship it because of concerns about the ongoing maintenance overhead, since we had other Python code generating GeoJSON that we would have still used - so we would have ended up maintaining multiple implementations.

simonw 5y ago

I worked on a Django project recently where we were generating GeoJSON from data in a PostgreSQL table. We investigated using tricks like this to generate the JSON directly from a SQL query - my notes here: https://til.simonwillison.net/postgresql/constructing-geojson-in-postgresql https://til.simonwillison.net/postgresql/constructing-geojso... The performance was a big enough improvement that, had we been bottlenecked on this particular operation, we would have shipped this (I'm afraid I don't have notes on benchmarks). We only decided not to ship it because of concerns about the ongoing maintenance overhead, since we had other Python code generating GeoJSON that we would have still used - so we would have ended up maintaining multiple implementations.