universal-symbols
v0.1.2
Published
A comprehensive collection of categorized symbols for expressive and formal programming
Downloads
58
Maintainers
Readme
Universal Utility Symbols
These are symbols you can use across your packages.
They use the symbol registry, but prefixed with a UUID. This ensures peer versions and future versions of this package won't cause issues.
If you want to be extra cautious, fork this repository and change the UUID prefix, then use the fork in your packages.
Here's how you use it:
npm install universal-symbolsimport { STATE } from 'universal-symbols';
const my_object = {
// Normal properties
a: 1,
b: 2,
c: 3,
// Symbol property
[STATE]: { current: 2 },
};TypeScript Support
This package includes full TypeScript support with automatically generated type declarations. All symbols are properly typed as unique symbol types, ensuring type safety when using them as object keys or method names.
import {
INTERNAL,
PRIVATE,
PROTECTED,
STATE,
CACHE,
EVENTS,
ACCESS_VIBES,
VIBES,
ALL_SYMBOLS
} from 'universal-symbols';
// Using symbols in interfaces
interface ConfigurableObject {
[INTERNAL]: any;
[PRIVATE]: any;
[PROTECTED]: any;
[STATE]: any;
[CACHE]: any;
}
// Using convenience collections
const accessLevels = ACCESS_VIBES;
const allVibes = VIBES;
const allSymbols = ALL_SYMBOLS;
// TypeScript will provide full IntelliSense and type checking
console.log(accessLevels.INTERNAL); // ✅ Type-safe
console.log(allVibes.STATE); // ✅ Type-safeThe TypeScript declarations are automatically generated from the symbols.json data file, ensuring they stay in sync with the JavaScript implementation.
For a comprehensive list of symbols, you can read the code file or the data file.
Symbol uses
Some of these symbols are classified as "vibes", while others are classified as "formal".
- formal symbols have associated descriptions of precisely what these symbols should be used for. If there is a situation where it is ambiguous whether or not one of these symbols is appropriate, that's considered to be a bug.
- vibes are symbols that were generated by AI. Their descriptions are intentionally vague. Use these as you see fit, just don't expect other packages to use them the same way.
Formal Symbols
OBJECT_EXTEND_FUNCTION_ENTRIES
This symbol is used when you have a function that is used to add members to an object, or create a new object with members added to it.
The function (fn) will have this symbol as a member
(fn[OBJECT_EXTEND_FUNCTION_ENTRIES]), and the associated value will
be an object containing entries that will be added.
const entries = { a: 1, b: 2 }
export const fn = function (o) {
return { ...o, ...entries };
}
fn[OBJECT_EXTEND_FUNCTION_ENTRIES] = entries;Why not just use const NAME = Symbol('name') or Symbol.for?
Using const NAME = Symbol('name') is fine if you're doing so within
a single package. If your package will be a dependency of other packages,
there may be different versions of your package installed which will have
their own instances of the symbol. As other packages pass them around you
will find that one NAME won't be equal to another NAME - oof!
Now, there are two solutions for this - one from javascript itself, the
other from how npm works.
Solution with Symbol.for
You can use Symbol.for to use the global symbol registry. All packages
that reference Symbol.for('CAT') will actually get the same Symbol.
This effectively doubles the set of infinity keys that objects can have. Yes, it sounds weird to "double" infinity, but you're sentient meat and that's pretty weird too.
Collisions are definitely possible here too. This repo prefixes symbol keys with a UUID to avoid this. However, packages using symbols from this package for a different purpose than how you're using those symbols could be in conflict as well. (that's why we suggest forking and changing the UUID if you want to be extra cautious)
Other packages in the Bikeshed Sanctuary organization will use these symbols in places where we want to accept javascript objects that may use any of the following:
- string keys
- non-registered symbols
- registered symbols
Having this particular set of symbols, prefixed with the UUID, makes a conflict much less likely, while still allowing references across symbols
Solution with peer dependencies
npm has an awareness of something called "peer dependencies". It's a
little difficult to understand but absolutely critical to understand
well. Here's an attempt at a clear explanation:
Consider this "happy" situation:
- You have a package "A" with dependencies "B" and "C"
- Dependencies "B" and "C" depend on "D", as a peer dependency
- "B" wants "D" at any major version
2 - "C" wants "D" at any version above
2.5.1 npmwill install one copy of "D" at version2.9
...and this "sad" situation:
- You have a package "A" with dependencies "B" and "C"
- Dependencies "B" and "C" depend on "D", as a peer dependency
- "B" wants "D" at exactly
2.6 - "C" wants "D" at any version above
2.7 npmwill throw an error and not install "D"
If we didn't use peer dependencies npm would install two copies
of "D" at different versions ant successfully complete. You might
then wonder why we would use peer dependencies at all if it just
makes it more likely a package will eventually break.
Here's why: all versions of package "D" at 2.x.x have compatible
interfaces, but opaque objects/symbols produced by that interface
may have internal behavior that differs between versions. Unless
you do integrations test between your package and every other
version of your package, including future versions when they're
released, you have absolutely no guarentees that anything will work
the way you expected. You could have 100% test coverage in your
repo and practically 0% coverage for peer use. This level of testing
is not feasible in most realistic scenarios.
Peer dependencies would also allow the same Symbol to be used. However,
if you don't have to worry about internal behavior destroying API
compatibility, you can avoid using peer dependencies and use this package
instead. You don't want to introduce peer dependencies if you actually
need the guarentees it's providing.
