Claude
b5ee699a28
None of the database calls in these three places were wrapped in try/catch, so any database hiccup (disk issue, lock contention, etc.) surfaced as an unhandled 500 with a useless generic message and no server-side log to explain why. Found this while testing every page's redirects end to end: a broken test setup on my end (an old server process left running against a deleted database file) produced exactly this failure mode, which would look identical to a real intermittent outage to an end user. Now: signup and profile-update return a clear "try again in a moment" message with the real error logged server-side; login logs the real error but still shows the existing generic "incorrect email or password" message to the client (matches current behavior, avoids leaking account existence). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VL5YUEf2B9LG4xdVUQwcSg