cat/CHANGELOG.md

74 lines
2.2 KiB
Markdown

Changelog
=========
Version 1.4.0
-------------
Adds documentation to a number of modules.
Adds an "equality principle" for categories and monads.
Prove that `IsMonad` is a mere proposition.
Provides the yoneda embedding without relying on the existence of a category of
categories. This is acheived by providing some of the data needed to make a ccc
out of the category of categories without actually having such a category.
Renames functors object map and arrow map to `omap` and `fmap`.
Prove that kleisli- and monoidal- monads are equivalent!
[WIP] Started working on the proofs for univalence for the category of sets and
the category of functors.
Version 1.3.0
-------------
Removed unused modules and streamlined things more: All specific categories are
in the namespace `Cat.Categories`.
Lemmas about categories are now in the appropriate record e.g. `IsCategory`.
Also changed how category reexports stuff.
Rename the module Properties to Yoneda - because that's all it talks about now.
Rename Opposite to opposite
Add documentation in Category-module
Formulation of monads in two ways; the "monoidal-" and "kleisli-" form.
WIP: Equivalence of these two formulations
Also use hSets in a few concrete categories rather than just pure `Set`.
Version 1.2.0
-------------
This version is mainly a huge refactor.
I've renamed
* `distrib` to `isDistributive`
* `arrowIsSet` to `arrowsAreSets`
* `ident` to `isIdentity`
* `assoc` to `isAssociative`
And added "type-synonyms" for all of these. Their names should now match their
type. So e.g. `isDistributive` has type `IsDistributive`.
I've also changed how names are exported in `Functor` to be in line with
`Category`.
Version 1.1.0
-------------
In this version categories have been refactored - there's now a notion of a raw
category, and a proper category which has the data (raw category) as well as
the laws.
Furthermore the type of arrows must be homotopy sets and they must satisfy univalence.
I've made a module `Cat.Wishlist` where I just postulate things that I hope to
implement upstream in `cubical`.
I have proven that `IsCategory` is a mere proposition.
I've also updated the category of sets to adhere to this new definition.