---
type: "post"
id: "datasette-1-0a43-adds-streaming-support-for-output-dffa1b"
slug: "datasette-1-0a43-adds-streaming-support-for-output-dffa1b"
title: "Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports"
description: "Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports, while moving the project toward its 1.0 milestone."
retrieval_nugget: "Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports, while moving the project toward its 1.0 milestone. It also fixes a privacy issue involving private stored queries and refines several interfaces. For data-application teams, the consequential change is a documented way for export renderers to emit table rows incrementally."
published_at: "2026-11-16"
updated_at: "2026-11-16"
record_date: "2026-11-16"
date_kind: "published_at"
topics: ["developer-tools"]
entities: ["Datasette","Simon Willison","Zihan Dai","Peter Olkhovets","Ilya Khalafi","Ankita Advitot"]
related_patterns: ["agent-ready-software-exposes-capabilities"]
novelty_label: "new_fact"
what_changed: "Datasette 1.0a43 adds streaming output renderers, fixes exposure of private stored-query metadata, and introduces several interface improvements."
prior_refs: ["https://newruntime.com/posts/datasette-1-0a41-analysis-0dd7a4/"]
source_url: "https://simonwillison.net/2026/Nov/15/datasette"
source_urls: ["https://simonwillison.net/2026/Nov/15/datasette"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:datasette-1-0a43-adds-streaming-support-for-output-dffa1b"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/datasette-1-0a43-adds-streaming-support-for-output-dffa1b/","markdown":"https://newruntime.com/posts/datasette-1-0a43-adds-streaming-support-for-output-dffa1b.md","json":"https://newruntime.com/posts/datasette-1-0a43-adds-streaming-support-for-output-dffa1b.json"}
markdown_url: "https://newruntime.com/posts/datasette-1-0a43-adds-streaming-support-for-output-dffa1b.md"
---

# Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports

## Retrieval answer

Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports, while moving the project toward its 1.0 milestone. It also fixes a privacy issue involving private stored queries and refines several interfaces. For data-application teams, the consequential change is a documented way for export renderers to emit table rows incrementally.

## Summary
Datasette 1.0a43 adds streaming support for output renderers, including built-in TSV and CSV exports, while moving the project toward its 1.0 milestone. It also fixes a privacy issue involving private stored queries and refines several interfaces. For data-application teams, the consequential change is a documented way for export renderers to emit table rows incrementally.

## What happened
Simon Willison’s announcement introduces Datasette 1.0a43 as another release on the path to Datasette 1.0. Willison says the 1.0 milestone is 53% complete and that he has been using Claude Fable 5.2 to help work through items on that milestone. The release’s largest 1.0-related item, in his description, is streaming support for output renderers.

The built-in TSV and CSV export mechanisms now use that streaming support. Plugins can also register that they are able to stream every row in a table in one go. The release adds a `Response.stream()` method for this purpose, alongside documentation for the relevant plugin hook. This makes the capability an extension point rather than an implementation detail confined to Datasette’s own export formats: a plugin author can declare a renderer capable of producing output as rows are handled.

Willison does not characterize this as a benchmarked performance release. The announcement gives no timings, memory figures, throughput numbers, or tested table-size ranges. Still, the shape of the interface matters. A renderer that can send rows progressively has a different operational model from one that must first construct a complete export. Datasette’s own CSV and TSV paths now use the newer model, while third-party renderers have a documented route to do the same.

The release also contains two security fixes. Willison says one was serious enough to receive a GitHub security advisory. Under the issue he describes, private stored queries could be exposed, revealing a query’s title and SQL. The exposed query could not be executed through that issue. This distinction limits the reported impact, but titles and SQL can still disclose internal naming, assumptions, data-model details, filters, or implementation logic that a team intended to keep private.

Several smaller changes accompany the renderer and security work. The JSON representation of a row page has been redesigned for 1.0. A new `datasette-copy-button` Web Component makes it possible to copy schema in several locations. JSON stored in columns can be pretty-printed on row pages, and JavaScript plugin authors gain a `DatasetteModal.confirm()` utility method. Willison also reports dozens of bug fixes and credits Zihan Dai, Peter Olkhovets, Ilya Khalafi, and Ankita Advitot for contributions.

Willison singles out a visual change as a personal favorite: CSS multi-page view transitions. The implementation enables cross-document navigation transitions for people who have not set a reduced-motion preference. It assigns transition names to facet results and table content, so those elements can visually carry across page navigations. Willison says this makes Datasette, which remains a multi-page application, feel more like a single-page application. He reports that it works in Safari and Chrome, and says Firefox support was expected in Firefox 162 on December 8, 2026 after the feature was no longer feature-flagged in Firefox Nightly.

## What the archive adds
New Runtime’s earlier coverage of [Datasette 1.0a41](https://newruntime.com/posts/datasette-1-0a41-analysis-0dd7a4/) on October 1, 2026 focused on runtime visibility and plugin-interface consistency. This is a new event in the same release sequence: 1.0a43 adds renderer streaming, changes built-in exports, and addresses the stored-query exposure rather than restating the earlier work.

## Trend signal
This release weakly supports [agent-ready-software-exposes-capabilities](https://newruntime.com/patterns/agent-ready-software-exposes-capabilities/). `Response.stream()` and the renderer hook expose a bounded, documented capability that plugins can adopt. Datasette is not presenting this as an agent feature, so the signal is not about autonomous agents; it is about software becoming easier to extend through explicit, composable interfaces.

## Why now
The timing is tied to Datasette’s approach to the 1.0 milestone. Willison says that milestone is 53% complete and identifies renderer streaming as the biggest 1.0 item in this release. The release therefore combines a foundational export-path change with interface cleanup, bug fixes, and a security response while the project is still converging on its first major stable version.

## Connections to other events
An earlier New Runtime note covered [Datasette Apps 0.3a0](https://newruntime.com/posts/release-0-3a0-datasette-datasette-apps-70d81b/) on November 8, 2026, which added permission-aware app references and stored-query controls for workflows using Datasette Agent. The private stored-query exposure fixed in 1.0a43 makes that broader direction more consequential: as query-based workflows acquire more controls, the confidentiality of the queries themselves remains an operational concern.

## Why it matters
Large exports are a routine stress point for data products. If a renderer can emit rows while processing them, a team may be able to avoid waiting for an entire export to be assembled before a response begins. That is an architectural opportunity, not a performance result claimed by the release: teams still need to test their own databases, deployment settings, client behavior, and custom renderers.

For Datasette plugin maintainers, the release provides a concrete integration surface. Instead of approximating streaming in an ad hoc renderer, they can use the documented output-renderer support and `Response.stream()` method. For operators, the security fix deserves separate attention. SQL text and query titles can reveal useful internal context even when execution is unavailable. A deployment using private stored queries should evaluate the linked advisory and release notes, then determine its upgrade path based on its installed version and use of that feature.

## What happens next
- **More Datasette 1.0 work will land around explicit plugin and runtime interfaces** (3 months, 70%). Watch for: milestone progress and release notes naming 1.0 items.
- **Plugin authors will begin testing streaming renderers against larger exports** (6 months, 60%). Watch for: plugin releases that adopt the output-renderer streaming hook.
- **The cross-document transition feature will gain a broader browser baseline if Firefox 162 ships support as expected** (3 months, 65%). Watch for: Firefox 162 release notes and Datasette browser-support updates.

## Limits and uncertainty
The announcement does not quantify streaming performance, memory behavior, supported table sizes, or behavior under slow clients and interrupted downloads. It does not say how many plugins will adopt the renderer capability. On security, it identifies exposure of private stored-query titles and SQL but does not state a severity score, affected-version range, deployment conditions, or whether exploitation was observed. The expected Firefox 162 support is a forecast reported by Willison, not a confirmed shipping outcome. Neither the security implications nor export behavior have been independently evaluated here.

## Skill to build now
If you maintain a Datasette plugin that exports table data, map its current response lifecycle this week. Identify whether it buffers all rows before writing output, then compare that design with the documented streaming renderer hook and `Response.stream()` method. Test with a representative large table and a deliberately slow client connection. Keep the test focused on observable behavior: when the first bytes arrive, whether pagination or ordering changes, how cancellation behaves, and whether the generated file remains valid.

## Reader takeaway
Datasette 1.0a43 is primarily an interface release: CSV and TSV exports now stream, and plugins gain a documented path to stream table rows too. It also fixes an exposure of private stored-query metadata and SQL. Treat the streaming feature as something to test in your own renderer and deployment, not as an unmeasured promise of universal performance gains.

## Open question
Will the new renderer interface become widely adopted by Datasette plugins, and will future releases publish measurements that show where streaming materially changes export reliability or resource use?

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Developer Tools](https://newruntime.com/topics/developer-tools.md): Explore the developer-tools topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)
