A tool update is only worth your attention if it changes your next decision.
If you mostly use ChatGPT or Claude and keep trying to figure out which new tools actually matter, Datasette 1.0a35 is easy to misread. You see "SQLite viewer," keep scrolling, and spend your attention on louder updates instead.
What the sources actually say: On June 16, version 1.0a34 added insert, edit, and delete for individual rows.[S002] On June 23, version 1.0a35 added create table, alter table, and a drop table button inside that flow.[S001] The JSON write API docs say those same actions can also run through POST-only API calls with tokens and granular permissions.[S003]
That is the whole turn. Row edits mean "change one record." Table create/alter/drop means "change the structure." Once the same jobs also exist behind an API, Datasette stops feeling like a glass window on data and starts feeling more like a small backend, meaning a tool that manages data instead of only showing it.
Don't judge an update by how many features shipped. Judge it by whether it changes your next decision.
My decision shift here is simple: stop filing Datasette under "viewer only." Start filing it under "small SQLite backend," at least from what the release notes and docs now show.
Boundary: I have not tested this on Mac, Linux, or Windows. This is a read of the June 16 and June 23 release notes plus the JSON API docs, not a production recommendation. And I would not stretch this into "general backend." The scope still looks like SQLite plus permissions, not an all-purpose system.
Share this with the person who still thinks Datasette is only a viewer. Would you trust it for a small internal tool yet?
#SQLite #IndieHackers #DeveloperTools #BuildInPublic