CHANGELOG.md
Rama 1.4 · réplica de lectura del repositorio original.
# Changelog
## 1.4.0
- Support _Table of Contents_ on documents.
- Introduced new `<ImageGrid>` documentation component.
## 1.3.0
- We've added support for Git backed versioning and multiple versions of a DDT.
- We've moved to a new GitHub organization.
- The docs have been extracted and moved to https://github.com/rundsk/website
- The JavaScript SDK has been extracted and moved to https://github.com/rundsk/js-sdk
- The example design system has been extracted and moved to https://github.com/rundsk/example-design-system
- The backend has been refactored and modularized.
- We've removed support for `go get` without additional initialization as this required us to
keep a comparitvely big go file around where assets had been inlined.
- We've moved from an internal Concourse pipeline to GitHub actions, which will run our test
and deploy the website. We are now able to automatically build and test packages for macOS.
- To control resource sharing we've introduced the `-allow-origin` flag, which
needs to be provided when starting DSK, if you like to allow origins from which
browsers can access API resources. A common scenario for this is when you have
multiple DSKs running and the frontend of DSK (a) wants to access the API of DSK
(b), i.e. for cross references.
- JSON and YAML assets can now be converted into the other type, just exchange the extension
to either `.json` or `.yaml`, when forming the URL to request the asset. To render the `foo.yaml`
asset as JSON use `/api/v1/tree/foo.json` instead of `/api/v1/tree/foo.yaml`. Both versions of the asset are also exposed in the “Assets” tab in the frontend.
- API responses for node assets that are images, now carry their dimensions in pixels. This information is also exposed in the “Assets” tab in the frontend.
- In addition to `@2x`, we've added `@x2` as a magic string, which when found in an image's file name, causes
it to be considered "retina".
- We've fixed some bugs where assets with an order number could not be loaded:
- Fix retrieval of assets, which have an order number.
- Fix retrieval of assets, which have decomposed file names.
- We now support HTML comments inside Markdown documents.
- The annotations for playgrounds have been improved.
- The `<Link>` component does now support providing a `target` property, i.e. `<Link target="new">`.
- The built-in frontend now supports captions for images, i.e. `<Image caption="Hello World">`.
- Introduced inline color card as `<Color>`, color cards are now named `<ColorCard>`.
- `<CodeBlock>` and markdown fenced code blocks now support code highlighting.
- Introduced new documentation components for embedding code examples: `<Asciinema>`, `<CodeSandbox>` and `<Glitch>`.
- We now display related aspects in built-in frontend.
## 1.2.0
- The frontend has been rewritten as a React-App, and redesigned
from scratch. It surfaces important features, like full text search and
just looks great.
- We've introduced documentation components, i.e. `<Banner>`, `<CodeBlock>`, or `<DoDont>`
to help you creating top class design documentation. This feature is currently limited
to Markdown documents.
- We now support freeform data under the `custom` key in `meta.yaml` (or `meta.json`). It
is made available through the API. The built-in frontend has been enhanced to display
the freeform meta data nicely. #62, #71, #72
- Certain aspects can now be configured through an optional configuration file.
- The window title is now formatted in a more search engine/human friendly way.
When an aspect has a description we use that to set a meta description.
- We now clean up Markdown documents better before they are indexed for search.
- Search results now return hit fragments, so highlighting can be implemented in the frontend.
- We unfiltered assets, so that these can be freely inlcuded or made available for
download in the frontend. It's now up to the frontend to decided whether an asset
should be made available for download. The next version of the API will not
include `downloads` on node results any more. We recommend to use the new `assets`
instead.
- The DDT is not a good place to store secrets and we want to be more clear about that. We've
removed special ignore rules and do not exclude aspects prefixed with `_`, `x-` or `x_`.
anymore. This clarifies a possible misunderstanding. Things you don't want to be accessed,
should not be stored inside the DDT.
- Each document now makes an array of (top-level) components available
through the API. We plan to expand the information that
is returned here for the next release.
- We fixed an issue where the backend didn't correctly respond to HTTP
requests. The backend now responds to missing API resources with a
404, this allows to use HEAD request to test for the existence of i.e.
a node.
- Our JavaScript SDK `frontend/js/dsk` is now a package, that we publish on npm.
- The JavaScript API client from the SDK, now supports pinging and
checking for the existence of a node or asset. In total three new
methods were added: `has()`, `config()` and `ping()`.
- The document transform from the JavaScript SDK, is used to parse documents
received from the backend and replaces occurences of components, with
actual component instances. The transformer has cleaned up and modified
to work with the new documentation components.
- Many of you wish to run DSK inside a docker container. We now prebuild and ship official
container imagee and make them available on
[on docker hub](https://cloud.docker.com/u/atelierdisko/repository/registry-1.docker.io/rundsk/dsk).
- Support for multiple languages has been removed, this unfortunately
didn't work out the way we expected to. We've also removed the `lang`
command line in favor of providing the language of the DDT via the new
optional configuration file. In the language is a property of the DDT
not the program.
- We've also removed support for using wide index with filter search,
all filter queries will use the narrow index.
- The `frontend` flag has been added, so the built-in frontend can be switched out without
compiling it in.
## 1.1.1
- Fix possible data race in repository lookup table, #58
- Fix issue where "Source" wasn't clickable in the built-in frontend, #66
- Fix incorrect date shown on downloads in the built-in frontend, #57
- Keep original query and fragment when making node URLs absolute, #65
## 1.1.0
The first minor release following the release of 1.0.0, featuring a
brandnew search build on top of the go native
[bleve](https://github.com/blevesearch/bleve)
and laying foundation by adding initial Git support.
- Add HTTP APIv2
- Introduce full search in APIv2, #44
- New and improved filter search in APIv2:
- now uses prefix matching instead of haystack/needle,
- slightly more immune to typos by using fuzzy matching,
- uses analyzers as full search does,
- supports `wide` mode,
- API responses now use keys named `nodes` instead of `urls` to return an
array of so-called _RefNodes_ with title and URL for each node. This
makes that part of the response uniform when compared to other API
responses.
- keywords are not searched anymore.
- Add initial Git support and read modified dates from it, if possible
- Replace go-bindata build time dependency with vfsgen, fixes #40 and #49
- Require Go v1.11, drop support for Go v1.9 and v1.10
- Use go modules in favor of of go dep
- Improve transliteration when creating node slugs
- Normalize strings when they are read from the filesystem, fixes #48.
- Deprecate `keywords` in meta data
- Rewrite built-in frontend
Thanks to [Zach Wegrzyniak](https://github.com/wegry/) for contributing to
the search implementation and edge-testing the Git foundations.
## 1.0.2 2018-09-06
- Add support for Go v1.11
## 1.0.1 2018-06-27
- Normalize strings when they are read from the filesystem, fixes #48.
## 1.0.0 2018-05-15
This is the first stable release of DSK, which we introduce in the
[release announcement post](https://atelierdisko.de/journal/post-167-dsk).