Licences in a Pack¶

This page documents the companion mod, not Lost Cities

Lost Cities reads no licence file and has no opinion about one. Everything here is a convention The Lost Cities - DevTool follows, described as of 3.0.0. Nothing here is a claim about Lost Cities' behaviour, so it carries no verification chips, for the same reason The DevTool Commands carries none.

TL;DR

Terms go in data/<namespace>/lostcities/license.txt, spelled exactly that way. An import reads it and reports what it found. An export carries what it found into the pack it compiles, one heading per namespace. A namespace that states nothing is reported as nothing found, which is not a finding of all rights reserved.

Where the file goes¶

	Path
Primary	`data/<namespace>/lostcities/license.txt`
Fallback	The pack root, beside `pack.mcmeta`

The primary path is not the root, and one case decides it. A pack living in kubejs/data/<namespace>/lostcities/ has no root of its own, and neither does one shipped inside a mod jar: the root there belongs to KubeJS or to the mod. Under data/ is the only place that travels with the assets in every shape a pack is distributed in.

Under data/, no other spelling works

Anything under data/ is a resource location, so every path segment is validated against [-._a-z0-9]+. LICENSE.txt, License.txt and licence.txt are therefore invalid paths, and the loader reports Invalid path in datapack ... ignoring rather than anything mentioning a licence.

A zip pack is not checked that way, so the identical file is readable there and not in the folder pack beside it. Nothing announces that difference. The lowercase American spelling is the one that works in both.

The pack root is not a resource location, so the fallback finds LICENSE.txt, COPYING and Licence.md there without trouble.

One statement per namespace¶

A namespace is one author's asset set, so one statement covers it. Per folder was considered and answers a question nobody has: terms differ between authors, not between a building and the palette it resolves through.

The file is free text and there is no schema. A .txt that somebody pastes MIT into is the version that gets adopted.

What an import shows¶

/lcdev import reports, for each namespace whose assets became plots:


How much	The first three non-blank lines, with leading blanks and centring stripped first, or Apache and the GPL show two empty lines and a run of spaces
The rest	Counted, not elided. "19 more lines" says whether this is a permissive licence or the whole GPL. An ellipsis says neither
Line width	Each line is cut to the chat width, because a licence written as one paragraph satisfies a line count and defeats it
File size	The read stops at 64 KB. This is a file from an untrusted pack, read on every import, to show three lines of

The namespaces reported are those whose assets became plots, not every namespace the walk touched. A city style or a world style is configuration; the buildings and the parts are the authored content, and they are what an export writes back out. That keeps the import message and the export behaviour saying the same thing.

What it says when it finds nothing¶

A missing file is weak evidence. Plenty of packs state their terms on a project page and ship nothing, and a legal determination printed by a tool is one that can be wrong with the author's name attached to it. So the message names both places it looked, then says how to proceed: copyright is automatic, so treat the work as all rights reserved unless the author says otherwise, and ask before redistributing.

Naming both paths matters where somebody does have a licence and it is not being found: that line is the only thing that says why.

Importing the mod's own pack reports nothing found

/lcdev import lostcities:standard is what everybody runs first, and it reports lostcities as stating nothing. That is literally true: the Lost Cities jar carries no licence file at its pack root or under data/lostcities/lostcities/.

It is also pessimistic. The project is MIT, stated in its repository. The message names no namespace as special, including that one, because a datapack can override any namespace and content under lostcities may be somebody else's overrides under different terms.

What an export carries¶

An export writes the statements it holds into lostcities/license.txt in the pack it compiles, one heading per namespace. A pack compiled out of somebody else's content does not ship with the credit stripped out.

/lcdev export <name> plot carries them too, and carries only that plot's, which is exactly the moment attribution should follow: lifting one building out of somebody's pack.

Three details of how that works:


The text is copied into the world at import	It lands under `<world>/lostcitiesdevtool/licences/`. Reading it again at export would be less state and would lose the statement exactly when it matters, because a workshop outlives the packs it was filled from and a pack removed from the world would take its author's terms with it, silently
A plot records where it came from	In the `source` setting. Without per-plot provenance a fragment lifted out of a three-pack workshop would ship all three authors' terms, and terms attached to work that is not in the file are a false statement rather than an over-cautious one
Carrying is idempotent	A pack compiled out of an imported pack that was itself compiled from an import would otherwise nest one heading inside another and label the first author's terms with the second author's namespace, accumulating every round trip. A notice carrying a notice passes its blocks through instead

/lcdev workshop clear confirm forgets the statements, after the backup pack it writes first has already carried them.

What this does not do¶

Prevention was never the goal. Nothing here can stop somebody reading files they already have, and a flag that refused an unlicensed import would bind only the people already being careful while leaving pack authors falsely confident. Attribution is the target.

A notice can be forged. Nothing distinguishes a notice this mod wrote from a file whose first line happens to match. Refusing to unwrap one would bring back the nesting that mislabels the first author's terms, which is the worse failure of the two.

The number of blocks shown is not capped, only what is inside each one. Reaching a problem would take about a hundred namespaces inside one world style.