Skip to content

nfty 0.5.0 — inspect reads a cooked Set

Choose a tag to compare

@corderro-artz corderro-artz released this 07 Sep 00:42
· 653 commits to main since this release

inspect reads a cooked Set

The reference page has listed .set among the kinds nfty inspect reads for a long time. It did
not — the extension table knew four kinds, so the command refused the one archive you are most
likely to have been handed by somebody else, with Unknown archive extension '.set'.

It reads one now, and reports what the run actually produced:

$ nfty inspect ./collection
Set: Chest Demo
  Assets: 500
  Seed: launch
  Unique DNA: required - every asset is distinct
  CookBook SHA-256: 4eee8f3d779ce313f4d226228c3e34d602cad606d8520cddfd42eef64fda3920
  Generator: nfty/1.0
Recipes (by id):
  chest               325   65.00%
  strongbox           175   35.00%
Traits (observed):
  Bands          Corners         47.50%
  ...

That is the complement to nfty stats, which reports what a CookBook's weights predict. Both
group their traits the same way, so the same trait is in the same place in each.

A Set folder works too, not just a packed .set. generate --out ./collection writes a folder
and --pack is optional, so inspect ./collection is the very next thing anyone types — and it used
to fail with "has no extension", which describes the mechanism rather than the situation.

--voxel is refused on a Set, as it already is on a Kitchen, and the message says where to run it
instead. A Set's images are output; partial alpha in one came from a variant in the CookBook that
made it, which is where it can still be changed.

Reading a Set costs the same whatever its size — the report comes from set.json alone, so a
10,000-asset collection inspects as fast as a two-asset one.

Smaller things

  • The GUI had grown its own .set extension check to route a remembered Set around the extension
    table. That is gone; both front-ends now ask the same question. Importing a .set by typing its
    name opens it instead of reporting it as a Kitchen.
  • distribution[].recipe in set.json was documented as the recipe's display name and has always
    held its id. Nothing rendered the field until now, so nobody was in a position to notice. The
    documentation was wrong, not the data — the report labels the column "by id" rather than quietly
    printing identifiers under a heading that reads like names.

Nothing about generation, the demo CookBook, or any file nfty writes has changed. Sets written by
earlier versions read exactly as before.

Suite: Cli 121 / Core 953 / App 641, zero warnings.