{{-- "We are updating" — one panel, used in both places it can appear. It showed up twice with two different faces: the console's own overlay in the app's light tokens while the application was still up, and then this, the 503 page, once the containers went down. Two designs for one event, one replacing the other mid-update. The styles are inline and self-contained rather than Tailwind classes, because the other place this renders is the 503 page — shown while the application is mid-deploy or freshly installed, exactly when the Vite manifest may not exist. A stylesheet reference there would throw instead of rendering the reassurance the visitor came for. Prefixed class names: this is injected into the console's own document, which has a stylesheet of its own, and `.badge` is not a name to claim globally. ── On the colours, because they were wrong twice ──────────────────────── Every value below is a copy of a token from resources/css/portal-tokens.css. They were not: the first version used Tailwind's slate ramp (#0f172a, #64748b, #e2e8f0), which is blue-tinted, against a product whose neutrals are warm. Even in daylight the panel read cooler than everything around it. And it carried a `prefers-color-scheme: dark` block, which nothing else in this product has. On a phone set to dark mode the update screen — and only the update screen — turned dark blue. Reported exactly that way. A dark mode for one page is not a dark mode; it is one page that does not match. --}}
{{-- The real mark, not a rounded orange tile. An empty brand square reads as an image that failed to load, and this panel is shown at the one moment somebody is already wondering whether something is broken. --}}
CluPilot Cloud
{{ __('updating.badge') }}

{{ __('updating.title') }}

{{ $body ?? __('updating.body') }}

{{-- Only in the console. A customer looking at the 503 has no use for a deployment step, and an operator watching a run that is stuck has no use for anything else: "läuft gerade" alone is what sends somebody to the shell. The markup is Alpine-driven and inert wherever Alpine is not — which is why it is behind a flag rather than always present. --}} @if ($detail ?? false)

{{ $starting ?? '' }}


            

        @endif
        
{{ __('updating.auto') }}


        {{-- Only when the server has stopped answering for long enough that
             this is no longer a restart. An always-present close button would
             be a lie — the deployment does not stop because somebody dismissed
             a panel — but being held behind one while the application is broken
             is worse. --}}
        @if ($detail ?? false)
            

                
{{ __('updating.stuck') }}

                
            

        @endif