Lifecycle event inconsistencies on enable and reload

[carlosjeurissen]

Introduction

Depending on the browser, an extension being switched from disabled to enabled and on reload results in different lifecycle events being fired.

Inconsistencies and demo

This is an overview:

	disabled -> enabled	reloaded
Chrome	-	onInstalled:update
Firefox	-	onInstalled:update
Safari	onInstalled:install	onInstalled:install
Orion	onStartup	onStartup

Demo extension:
https://github.com/carlosjeurissen/webext-tech-demo-extensions/tree/main/demos/background-lifetime-events

To clarify, orion does not fire onInstalled, yet magically fires onStartup. See bugreport:
https://orionfeedback.org/d/12603-when-extensions-get-reloaded-and-changed-from-disabled-to-enabled-it-fires-onstartup-instead-of-oninstalled

Next steps

Finding agreement on what should happen in these situations would be very valuable before browsers implement the proposed onEnabled and onExtensionLoaded events so at least these events will behave consistent across browsers.

See: #353 and https://github.com/w3c/webextensions/blob/main/proposals/runtime_on_load_on_enabled_events.md

In general seems on reload, onInstalled:update is most intuitive. If removing onInstalled:install on enabled, a potential alternative could be to add onInstalled:enabled, while not native to onInstalled, it could be a nice middle ground.

[nasreenkhannam]

Hi,I am interested in working on this issue.Could you please assign it to me and guide me to the relevant files?

[dotproto]

@carlosjeurissen, what was your testing procedure for the reload scenario? There are at least two strategies you could use: the "reload" button in the extension management UI and browser.runtime.reload(). Some browsers also allow developers to Ctrl/Cmd+R in developer tools to reload an extension. I think (but am not sure) that this uses the same reload mechanism as the first option I mentioned.

[carlosjeurissen]

@dotproto from my testing, there is no difference in outcomes between reloading with the extension management UI and reloading with browser.runtime.reload().

As for Ctrl+R, just tested this in Firefox, and it behaves the same as the others.

[nasreenkhannam]

I agree that this should be addressed at the spec level rather than by individual browser behavior.

From a developer perspective, overloading onInstalled.install for enable actions creates ambiguity. A clearer model would be:

• Reload → onInstalled with reason "update"
• Disabled → enabled → a dedicated onEnabled event

This separation would make lifecycle handling more predictable across browsers and reduce the need for defensive workarounds.