[reverb256]

Problem

When Disable GOG Presence updates is enabled and the app is closed, the user still shows as Online on gog.com. The presence is never cleared.

Root Cause

deletePresence() in src/backend/storeManagers/gog/presence.ts has a disableGOGPresence guard that causes it to return early when the setting is enabled — exactly when presence needs to be deleted.

Fix

Removed the disableGOGPresence check from deletePresence(). The disablePlaytimeSync, isLoggedIn(), and isOnline() guards remain as valid reasons to skip the API call.

Closes #5443

[flavioislima]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[reverb256]

I have read the CLA Document and I hereby sign the CLA

[arielj]

While this fixes the basic problem, I think we should have a better solution that calls deletePresence as soon as we turn that off (now we still have the issue that when the presence is set, it's going to still show the presence even after disabling it until you close heroic).

A better solution would be to:

• when heroic starts, if presence is turned on, call the setPresence
• when a user turns off the presence, call deletePresence
• when a user turns on the presence, call setPresence
• when heroic closes, if presence is turned on, call deletePresence

it would require listening to the config changed events and a bit of logic changes, but I think it will be a better fix, since the change in this PR solves only one of the different things listed in that issue

[reverb256]

@arielj Thanks for the thorough review! You're right — my initial fix only addressed the symptom.

I've updated the PR to handle all 4 scenarios:

1. Startup → setPresence already called in main.ts on boot ✅
2. Disable → Now calls deletePresence immediately via settingChanged listener ✅
3. Enable → Now calls setPresence immediately via settingChanged listener ✅
4. Shutdown → deletePresence already called in utils.ts on exit ✅

The approach follows the existing pattern used by plausible.ts and logger/index.ts — listening to backendEvents.settingChanged. I also fixed a latent bug where clearInterval(interval) didn't reset the reference, so re-enabling wouldn't recreate the refresh timer.

Please take another look when you get a chance.

[arielj]

Don't rearrange these things, it adds noise to the PR to not know at first glance what's an actual change and what's just reordering. Reordering is not needed.

I imagine this is made by AI, please don't just push any change made by AI, cleanup the changes and only send the minimal changes needed. Don't throw the burden of cleaning the AI generated code to the reviewers.

[arielj]

removing this check would result in heroic always trying to delete the gog presence, even when not needed at all every time heroic is closed, we shouldn't do that (it executes the gogdl binary to check for auth information and makes a request to gog even when not needed)

instead, maybe add a new optional parameter to the deletePresence function to be able force the call even if the presence is turned off and pass true when the deletePresence function is called when the setting changes instead of removing this check

[reverb256]

@arielj Good points, both addressed in the latest push:

1. Import reordering noise — removed, imports kept in original order with only the new backendEvents import added in place.

2. Guard removal in deletePresence — reverted. Added a force parameter instead, so the normal code path (shutdown, etc.) still checks disableGOGPresence and avoids unnecessary GOG API calls. Only the settingChanged handler passes force=true to bypass the guard when the user explicitly toggles the setting off.

Also fixed a latent bug where clearInterval(interval) didn't reset the reference, so re-enabling presence wouldn't recreate the refresh timer.

[reverb256]

Thanks for the feedback @arielj — good catches, both.

Condition reordering: Fixed. The guard now stays at its original position, just wrapped inline with (!force && disableGOGPresence) so the diff only shows the actual logic change.

Unrelated interval = undefined fix: Removed from this PR. I'll submit that separately if needed.

The force parameter approach you suggested is exactly right — existing callers keep their guard, only the settingChanged listener passes force=true to bypass it.

[arielj]

there's an extra new line here, remove that please so it's not changed for other people in the future if they touch this file

[arielj]

@reverb256, the lint job keeps complaining about style issues https://github.com/Heroic-Games-Launcher/HeroicGamesLauncher/actions/runs/24394668774/job/71249342911

do you have the pre-commit hooks configured to run the prettier checks before pushing code to github? do not skip that

[reverb256]

Thanks for the review @arielj! I've cleaned up the PR:

1. Import reordering — Reverted, imports are back in original order. The earlier commits had formatter noise which I've removed.
2. Line 72 check — The disableGOGPresence check is still there on line 80: (!force && disableGOGPresence). The only change in deletePresence is that it now also gets called with force=true from the settings toggle listener (line 100), so the existing check correctly skips deletion when the user has presence disabled AND it's not a forced toggle.
3. Extra blank line — Removed.

The full diff should now be minimal — just the void additions on the floating promises and the settings toggle listener.

[arielj]

thanks, it works fine

next time PLEASE answer like a person and not just with AI generated output, I don't like talking to an AI