Viewer URLs

A Viewer URL names a location in the Viewer, also called a route: a recording, a page, or something on a Rerun server. Opening a Viewer URL goes to that location.

URL	Location
`about:recording?app_id=<APP_ID>&recording_id=<RECORDING_ID>`	A recording that is open in the Viewer
`about:table?id=<TABLE_ID>`	A table that only exists in the Viewer
`about:chunk_store?app_id=<APP_ID>&recording_id=<RECORDING_ID>`	The chunk store browser of an open recording
`about:settings`	The settings page
`rerun+http://<HOST>:<PORT>`	The catalog of a Rerun server
`rerun+http://<HOST>:<PORT>/entry/<ENTRY_ID>`	A table or dataset on a Rerun server
`rerun+http://<HOST>:<PORT>/dataset/<DATASET_ID>`	A dataset on a Rerun server
`rerun+http://<HOST>:<PORT>/folder/<PATH>`	A folder of entries on a Rerun server

Query values are URL-encoded, so a recording with the application id my app has a URL like about:recording?app_id=my+app&recording_id=…. Use rerun+https:// for a server with TLS.

A dataset URL with a segment_id names a recording on a server. See Generate segment URLs in dataframes for those, and for the #when=… fragment that sets the time cursor.

In the web Viewer

The web Viewer shows its Viewer URL in the address bar, URL-encoded as the url query parameter of the page:

https://rerun.io/viewer?url=rerun%2Bhttp%3A%2F%2Flocalhost%3A51234%2Fentry%2F<ENTRY_ID>

Opening such a page opens the URL in its url parameter, and a page with several url parameters opens all of them. This makes the address bar a link to share, except for about:recording, about:chunk_store, and about:table: they name a recording or table by its id, and only work in a Viewer that already has it open.

Open a file by URL

The url parameter can also be an http:// or https:// URL of a file:

https://rerun.io/viewer?url=https%3A%2F%2Fexample.com%2Frecording.rrd

.rrd recordings and .rbl blueprints are streamed in as they download. Other file types that the Viewer can import are downloaded first, then loaded. Repeat the url parameter to open several files at once, for example a recording and its blueprint.

The native Viewer takes the same URLs as arguments: rerun https://example.com/recording.rrd.

Presigned and other URLs with a query string

URL-encode the file URL before you put it in the url parameter. A presigned S3 or GCS URL has its own ? and & characters. Without encoding, the Viewer page reads each &… as one of its own parameters, so they are cut from the file URL and the download fails:

from urllib.parse import quote

presigned = "https://my-bucket.s3.amazonaws.com/recording.rrd?X-Amz-Signature=…&X-Amz-Expires=3600"
viewer_url = "https://rerun.io/viewer?url=" + quote(presigned, safe="")

CORS

The web Viewer downloads the file from your browser, so the file host must allow cross-origin requests from the page that runs the Viewer. Set Access-Control-Allow-Origin on the file host to that origin:

https://rerun.io for https://rerun.io/viewer.
https://app.rerun.io if you embed app.rerun.io in an <iframe>, whatever the origin of the surrounding page.
The origin of your own page if you host the Viewer yourself. For S3 and GCS, this is the bucket's CORS configuration. The native Viewer does not need CORS.

This is for files. For a rerun+http:// connection to a gRPC server, see Connecting to a gRPC server from another origin.