fix(studio): audit batch — debug.echo install, copyable errors, fast Diagnose

From a live-test audit:
- Install via the flow-editor 'Fix' sent 'debug.echo@^0' (version
  constraint included) as the install source; the hub resolves by bare
  name so it missed ('no store entry for debug.echo@^0'). Strip the
  @<constraint> like the Store page does — debug.echo now installs.
- Errors the operator could not copy: route the daemon-start failure
  (its stderr!), the flow-editor install failure and the federation
  issue failure through showFaiErrorSnack / a new showFaiProcessError
  helper, so every error is selectable, one-tap copyable and logged.
- Diagnose page opened slowly because doctor() Future.wait-ed on a live
  manifest fetch (8s server timeout); cap that one check at 2s so the
  page no longer waits on the network.

Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
This commit is contained in:
flemming-it 2026-06-18 13:02:21 +02:00
parent ee4a0f6c62
commit 5f56abcf5f
5 changed files with 85 additions and 30 deletions

View file

@ -902,7 +902,18 @@ class HubService {
.catchError((_) => <PendingApprovalEntry>[]),
_client.verifyEventChain().catchError((_) => VerifyEventChainResponse()),
_client.listServices().catchError((_) => <DeclaredService>[]),
_client.checkUpdate().catchError((_) => CheckUpdateResponse()),
// checkUpdate does a live manifest fetch (server-side 8s timeout).
// The other five are local-socket RPCs that return in ms, so cap
// this one tightly otherwise it alone gates the whole Diagnose
// page open. A timeout reads as "release host unreachable", which
// the UI already renders as an offline banner, not an error.
_client
.checkUpdate()
.timeout(
const Duration(seconds: 2),
onTimeout: () => CheckUpdateResponse(),
)
.catchError((_) => CheckUpdateResponse()),
_client.daemonPaths().catchError((_) => DaemonPathsResponse()),
]);