Logs and telemetry (apiVersion 6)

Three ways to find out what happens to your plugin on someone else's device, from the most local to the one that leaves the device. All from "apiVersion": 6 (Kino 0.9.50); below it the fields are ignored.

What	Where you see it	Who turns it on
kino.log	adb logcat	always
Modo debug	error panel on screen, "Registro" page	the person, with the switch every plugin has ("debug": true makes it on by default)
"telemetry": true / "verbose"	the maintainers' error tracker	you ask and the person approves (no switch to turn it off yet)

Logcat

kino.log and console.* go to adb logcat under the tag KinoPlugin; in a debug build of Kino, or in any build while your plugin's Modo debug switch is on, under KinoPlugin/<your id>. Kino's own playback metrics go under the tag KinoPlay in those same cases, so

adb logcat -s KinoPlay KinoPlugin/<your id>

shows Kino's lines and yours together.

Modo debug: errors on screen and the Registro page

Since Kino 0.9.50 every installed plugin (yours, a generated Stremio addon, a Nuvio scraper) has a "Modo debug" switch in its own tab in Ajustes (Ajustes ▸ your plugin's name, phone and TV), with the line "Muestra los errores de este plugin en pantalla y guarda un registro que puedes compartir con su autor." You don't have to do anything for it to be there. While it is on:

• every failed call of your plugin shows a panel on screen with the function, the error code, the technical message, the JavaScript stack and your last kino.log lines;
• the tab gets a "Ver registro" button that opens the Registro: the last 200 events (your lines, the failures and each playback's kino:play … lines), with "Copiar registro" and, on the phone, "Compartir registro".

While it is off, nothing is shown and nothing is kept; turning it off clears the Registro.

This is how a person sends you what went wrong. When someone reports a problem, ask them to turn on Modo debug for your plugin in Ajustes, repeat what failed, and send you a screenshot of the panel or the copied/shared Registro.

"debug": true in your manifest only sets the default. With it, the switch starts on for everyone who installs the plugin (they still see your panels until they turn it off); without it (or with false), the switch starts off. Use true while you develop, or for a test build you hand to testers; for a plugin you publish to everyone, leave it out and let each person turn it on when they need to send you a report. validate.mjs adds a note when it is true. Once a person touches the switch, their choice is kept across updates (an update that adds or drops debug only moves the default of people who never touched it) and syncs to their other devices. Any value other than true or false is refused with "El campo \"debug\" debe ser true o false".

The Registro is kept in a private file on the device, so it survives a restart; it is never synced or backed up, and it is deleted when the switch goes off or the plugin is uninstalled. Secrets, and the person's own passwords, are masked there like everywhere else, so a shared Registro carries neither.

telemetry: your lines reach the error tracker

From Kino 0.9.50 only a plugin that declares "telemetry" sends its kino.log lines when a call fails, recommended or not, whatever repository it comes from. For any other plugin (a converted Nuvio scraper included) Kino only notes that the call failed: your id and version, the function and the kind of failure, never a log line.

• Consent. The consent sheet says "Comparte registros de errores con Kino para corregir fallas". An update that newly declares it waits for the person's approval, like a new host ("Actualización disponible — requiere tu aprobación").
• No switch yet. For now, while plugins are being stabilized, a declared plugin's lines are always sent. A later Kino build adds an "Enviar registros de errores" switch to your plugin's tab in Ajustes, on by default, that the person can turn off on each device; then nothing leaves while it is off.
• What is sent. When a call fails (it throws, times out, returns something unusable, including sign, settingsStatus, action and validateSettings), the lines it logged during that call (the last 30, each at most 300 characters, 2 KB in all) as plugin_log, tagged with your plugin's id and version. At most one report per function and kind of failure an hour. A call that succeeds sends nothing.
• What is removed before it leaves. URLs, hostnames, IPs, e-mails, long ids, long hex/base64 runs, credential-shaped text, the person's setting values and the text of their search or title. Still: log what happened (a status, a step, a count), never what the person typed, a secret or a setting's value.
• Any value other than true, false or "verbose" is refused with "El campo \"telemetry\" debe ser true, false o \"verbose\"".

"verbose"

"telemetry": "verbose" shares everything true does plus the playback metrics of a sample (a quarter) of the plays that went well, live and cast problem reports, and edge cases (a re-resolve, a failover, a decoder switch, a sign timeout, migrate results, settings synced from another device). At most 60 events per plugin until Kino restarts and one a minute per area. Its consent line is "Comparte registros detallados de reproducción y errores con Kino para corregir fallas"; an update from true (or nothing) to "verbose" waits for the person's approval, while "verbose" to true applies silently.

kino.log.report: a degraded result

kino.log.report(...args) (with telemetry) writes a line like kino.log and also tells the error tracker that your plugin served a degraded result even though the call worked: it fell back to a shared account, used a backup source, trimmed a list.

kino.log.report("myplugin:session", "shared_fallback", "tries=2");   // area myplugin:session

• The line's first word names the area, and must be a namespaced word of lowercase letters, digits, _ and : with at least one _ or :, up to 24 characters. Any other first word (a bare word, anything with a dot, @ or /, or one that holds one of the person's values) is filed as other.
• At most one report per plugin and area an hour and 3 per plugin until Kino restarts (and 10 in all, every plugin together, in that same run of Kino), sent as a warning. That cap leaves room for real failures, which have caps of their own.
• The whole line is scrubbed like any log line, the text of the call running at the time included.
• Without telemetry (or, once that switch exists, with it off) it is just a log line.
• Report what happened in codes and counts, never values that came from a response.

Playback metrics and problem reports

Kino measures every playback of a plugin stream (a film, a chapter, a live channel) and every cast of one to a TV, with no code in your plugin. Each playback gets one record: how long resolve took, the time to the first frame (the zap time for a channel), how the player reached the stream (direct, proxy, signed_proxy, remux), the video decoder and its switches, resolution and bitrate changes, rebuffers and the time spent stalled, player errors by class with their HTTP status, retries and their reason (expired, conflict, network, cut…), and for a request-signed stream sign p50/p95/max and timeouts per playlist and segment, fetch p50/p95 and errors per host index (0 = the stream's own host, 1… = its alternateHosts).

Detectors watch for what a viewer feels: no first frame in 10 s, a frozen picture, long stalls, dropped-frame bursts, audio underruns or sink errors, audio and video drifting apart, the audio track lost, decoder errors, falling behind the live window, HTTP errors per segment class; and for a cast the receiver's load timeouts, errors and idle reasons, a remux that stopped, the TV starting far from the phone's position, and a session that dropped. Nothing they write holds a URL, a host, a token or anything the person typed: only numbers and Kino's own words.

Where it goes:

• your plugin's Registro (while its Modo debug switch is on), one kino:play … line per milestone and a summary line;
• logcat under the tag KinoPlay in a debug build of Kino, or in any build while that switch is on;
• the error tracker, only with telemetry (and, once that switch exists, while the person leaves it on): with true, one summary per playback that ended on an error the person saw; with "verbose", also a quarter of the playbacks that went well, and one event per problem or edge case (at most 60 per plugin until Kino restarts, one a minute per area). Failure events carry the Kotlin exception's stack and, when your script threw, its own stack frames (at fn (plugin.js:12:5), frames only).