It never has: setDevelopment() is commented out in main.tsx, so the dev server
falls through to its /v1 proxy at api.storyteller.ai. Both bootstrap epilogues
and the seed script now say so and print the VITE_DEV_API_HOST needed to
actually reach the local backend.
Brings up a fully local storyteller-web: portable MySQL 8.4 and Redis under a
gitignored .devstack/ (plain user processes -- no admin, no services, no WSL,
no Docker), all 281 migrations, role/badge seeds, and a demo login with
credits. Windows and Linux variants, plus a doctor that reports what is and
isn't healthy.
Scripts and docs only. Nothing here is compiled into any binary, and no
application behaviour changes: the backend it starts is the one that ships.
Integrations that cannot be provided locally -- object storage, generation
providers, email -- get placeholder credentials, so those endpoints fail when
exercised. Swap in a real dev credential for the one you are working on. The
doc records exactly which areas that leaves working.
An earlier revision closed those gaps with development-only behaviour inside
storyteller-web (DEV_FAKE_GENERATION, a filesystem bucket mode, a CDN_BASE_URL
override). That is deliberately not here: a switch that exists in the shipped
binary can be flipped. Frontend work that wants a populated, working stack
should use frontend/apps/fake-storyteller-web instead.
We need to remove irrelevant FakeYou records that do not impact the
ArtCraft product
- 87M irrelevant media file records (no relation to ArtCraft)
- 105M jobs records (which we purge occasionally anyway)
- lots of other files: audit logs, old/irrelevant stripe event logs,
feed data, etc. etc.
Two reasons for this:
- super brittle when pinned to non-main commit
e1a70095c4409478c1ded499fb8dd4afa540fb90
- I was recently stuck on a flight without good wifi and setting up a
new devenv. Cargo was totally unable to download the remote github
dependency, but was fine downloading the entire artcraft repo. Seems bad
to link to other repos that Cargo can't quickly pull on low bandwidth.
We'll need to slim this down later and eventually get back on main
branch.
Tanish Manku