This section goes over the setup of the repo for development.
npm install -g pnpm
- Install dependencies
pnpm install
- Build the dependencies
pnpm build
- (Optional) Install Playwright browsers for UI testing
npx playwright install
- Start the build in watch mode to automatically rebuild on save
pnpm run watch
If you are not at the root of the repo you have to use -w
option to specify you want to run the command for the workspace. pnpm -w <command>
.
Those commands can be run on the workspace or in a specific package(cd ./packages/<pkg>
).
Command | Description |
---|---|
pnpm build |
Build |
pnpm test |
Test |
pnpm test:watch |
Run test in watch mode(only when inside a package) |
pnpm watch |
Build in watch mode, Starting this command will rebuild the typescript files on save. |
pnpm clean |
Clean, sometimes there are ghost files left in the dist folder (common when renaming or deleting a TypeScript file), running this will get a clean state. |
pnpm format |
Format |
pnpm format:check |
Validate files are formatted |
pnpm gen-extern-signature |
Regenerate TypeScript signature for decorators(except compiler) |
pnpm change add |
Add a change description |
pnpm lint |
Run linters |
pnpm lint:fix |
Fix autofixable issues |
pnpm regen-samples |
Regen the samples(when the samples test fail) |
pnpm regen-docs |
Regen the reference docs |
Tests sometimes log extra info using logVerboseTestOutput
To see
this output on the command line, set environment variable
TYPESPEC_VERBOSE_TEST_OUTPUT=true.
For the compiler you will need to run it manually or run the whole workspace build. This is because for the tool to run it needs the compiler to build first.
- Vitest Test Explorer: Run tests from the IDE.
- Prettier: Automatically keep code formatted correctly on save.
- ESLint: Show eslint errors in warnings in UI.
- Code Spell Checker: Show spell check errors in document.
Always open the root of the repo as the workspace. Things are setup to allow easy development across packages rather than opening one package at a time in the IDE.
- File -> Open Workspace, select root folder where the TypeSpec repo was cloned
- Or run
code /path/to/repo/root
on the command line
- Terminal -> Run Build Task (
Ctrl+Shift+B
)
This will setup a an incremental watching build for the whole repo. From there on, your changes will be built whenever you save.
Problems will be reported in the Problems pane automatically and the Terminal pane will have three parallel watch tasks running:
watch-source
: tsc process that recompile on TypeScript changeswatch-spec
: process that regenerates spec.html when spec.emu.html changeswatch-tmlanguage
: process that regenerates typespec.tmlanguage when tmlanguage.ts changes
# Run all the tests
pnpm test
# Run in a specific package tests in watch mode
pnpm test:watch
There are several "Run and Debug" tasks set up. Click on the Run and Debug icon on the sidebar, pick one from its down, and press F5 to debug the last one you chose.
- VS Code Extension: This will run and debug an experimental instance of VS Code with the TypeSpec extension for VS Code and TypeSpec language server running live with any of your changes. It will attach to both the VS Code client process and the language server process automatically.
- Compile Scratch: Use this to debug compiling
packages/samples/scratch/*.tsp
. The TypeSpec source code in that folder is excluded from source control by design. Create TypeSpec files there to experiment and debug how the compiler reacts. - Compile Scratch (nostdlib): Same as above, but skips parsing and evaluating the TypeSpec standard library. Sometimes it's easier to
- Attach to Default Port: Use this to attach to a manually run
node --debug
command. - Attach to Language Server: Use this to attach to the language server process of an already running client process. Useful if you want to debug the language server in VS Code while debugging the VS client in VS.
- Regenerate .tmlanguage: This runs the code that produces the typespec.tmlanguage file that provides syntax highlighting of TypeSpec in VS and VS Code. Select this to debug its build process.
Install Visual Studio 17.0 or later. It is not currently possible to build the VS extension without it, and of course you'll need Visual Studio to run and debug the Visual Studio extension.
See the command line build steps above. If you have VS installed, the VS extension will be included in your command line full repo builds automatically.
If you do not have VS installed the command line build steps above will simply skip building the VS extension and only build the VS Code extension.
- Open packages/typespec-vs/Microsoft.TypeSpec.VisualStudio.sln in Visual Studio
- Build -> Build solution (
Ctrl+Shift+B
)
Unlike TypeScript in VS Code above, this is not a watching build, but it is relatively fast to run. Press Ctrl+Shift+B again to build any changes after you make them.
- Click on the play icon in the toolbar or press
F5
This will run and debug an experimental instance of VS with a version of the TypeSpec extension for VS Code running live with any of your changes to the extension or the TypeSpec language server.
The VS debugger will attach only to the VS client process. Use "Attach to Language Server" described above to debug the language server in VS Code.
pnpm dogfood
This will globally install the @typespec/compiler package, putting your
build of typespec
on PATH, and install the VS Code extension if VS Code
is installed.
Note the important difference between this and the steps to run and
debug the VS Code extension above: the dogfood
command installs the
TypeSpec extension with your changes in regular, non-experimental instance
of VS Code, meaning you will have it always, and not only when running
the debug steps above. This is exactly like using tsp vscode install
,
only instead of downloading the latest release, it uses a build with your
changes applied.
There is no automatic dogfood
process for installing the VS
extension non-experimentally, but if you build the typespec-vs project from
the command line following the steps above, or build its Release
configuration in Visual Studio, then you can install it by
double-clicking on packages/typespec-vs/Microsoft.TypeSpec.VisualStudio.vsix
that gets produced.
Go to packages/website
and run the command:
pnpm start
For contributors of the repo the build will trigger automatically but for other's forks it will need a manual trigger from a contributor. As a contributor you can run the following command to trigger the build and create a TypeSpec playground link for this PR.
/azp run typespec - pr tools
Each team might use their own way of triaging issues however figuring out the area is a common process.
flowchart TD
classDef bot fill:#69f
classDef user fill:#9c6
subgraph "Legend"
a([User action])
b[\Automation detect/]:::bot
c[Automation action]:::bot
d{{state}}
end
subgraph "Issue creation"
select_template([User select template])
select_template --> bug_template
select_template --> feature_template
select_template --> plain
bug_template([Bug]) --> add_bug_label[π€ add `bug` label]:::bot --> start
feature_template([Feature]) --> add_feature_label["π€ add `feature` label"]:::bot --> start
plain([Plain]) --> start
start{{"β
Issue created"}}
end
subgraph "Area triage"
auto-triage[\π€ Detect if issue has area checkbox/]:::bot
add-needs-area[π€ label 'needs-area']:::bot
auto-area-label["π€ label '{area}'"]:::bot
add-area-label(["Issue is labelled with {area}"])
remove-needs-area[π€ Remove 'needs-area']:::bot
end
subgraph "Team triage"
team-triage{{Issue Labeled with Team area}}
team-triage -- wrong area --> remove_area([Remove area label])
team-triage -- correct area --> triage([Triage/add to project])
end
start --> auto-triage
auto-triage -- no --> add-needs-area
auto-triage -- yes --> auto-area-label
auto-area-label --> remove-needs-area
add-needs-area --> add-area-label
add-area-label --> remove-needs-area
remove-needs-area ---> team-triage
remove_area --> add-needs-area
TypeSpec repo use labels to help categorize and manage issues and PRs. The following is a list of labels and their descriptions.
Area of the codebase
Name | Color | Description |
---|---|---|
compiler:core |
#453261 | Issues for @typespec/compiler |
compiler:emitter-framework |
#453261 | Issues for the emitter framework |
ide |
#846da1 | Issues for VS, VSCode, Monaco, etc. |
lib:http |
#c7aee6 | |
lib:openapi |
#c7aee6 | |
lib:rest |
#c7aee6 | |
lib:versioning |
#c7aee6 | |
meta:blog |
#007dc8 | Blog updates |
meta:website |
#007dc8 | TypeSpec.io updates |
tspd |
#004185 | Issues for the tspd tool |
emitter:client:csharp |
#e1b300 | Issue for the C# client emitter: @typespec/http-client-csharp |
emitter:client:java |
#e1b300 | Issue for the Java client emitter: @typespec/http-client-java |
emitter:client:python |
#e1b300 | Issue for the Python client emitter: @typespec/http-client-python |
emitter:graphql |
#957300 | Issues for @typespec/graphql emitter |
emitter:json-schema |
#957300 | |
emitter:protobuf |
#957300 | The protobuf emitter |
emitter:openapi3 |
#957300 | Issues for @typespec/openapi3 emitter |
openapi3:converter |
#957300 | Issues for @typespec/openapi3 openapi to typespec converter |
emitter:service:csharp |
#967200 | |
emitter:service:js |
#967200 | |
emitter:service:java |
#967200 | Issue for Java service emitter |
eng |
#65bfff | |
ui:playground |
#3256a8 | |
ui:type-graph-viewer |
#3256a8 |
Issue kinds
Name | Color | Description |
---|---|---|
bug |
#d93f0b | Something isn't working |
feature |
#cccccc | New feature or request |
docs |
#cccccc | Improvements or additions to documentation |
epic |
#cccccc |
Labels around annotating issues and PR if they contain breaking change or deprecation
Name | Color | Description |
---|---|---|
breaking-change |
#B60205 | A change that might cause specs or code to break |
deprecation |
#760205 | A previously supported feature will now report a warning and eventually be removed |
Design issue management
Name | Color | Description |
---|---|---|
design:accepted |
#1a4421 | Proposal for design has been discussed and accepted. |
design:needed |
#96c499 | A design request has been raised that needs a proposal |
design:proposed |
#56815a | Proposal has been added and ready for discussion |
Process labels
Name | Color | Description |
---|---|---|
needs-area |
#ffffff | |
needs-info |
#ffffff | Mark an issue that needs reply from the author or it will be closed automatically |
stale |
#ffffff | Mark a PR that hasn't been recently updated and will be closed. |
triaged:core |
#5319e7 |
Misc labels
Name | Color | Description |
---|---|---|
Client Emitter Migration |
#FD92F0 | |
good first issue |
#7057ff | Good for newcomers |
mq |
#0969da | Good candidate for MQ |
Labels are configured in eng/common/config/labels.ts
. To update labels, edit this file and run pnpm sync-labels
.
If you create a new label in github UI without updating the labels.ts
file, it WILL be automatically removed
The various language emitters will live in the in the repo under the following directory structure
packages/{protocol}-{client|server}-{language}
- Contains the@typespec/{protocol}-{client|server}-{language}
package which is intended to be consumed by customers in their tsconfig.yaml file.packages/{protocol}-{client|server}-{language}-generator
- Contains the@typespec/{protocol}-{client|server}-{language}-generator
package which is the backend implementation of for a given emitter and usually contains code languages such as .NET, Python, or Java. This package is only intended to be used as a dependency of the root emitter package.packages/{protocol}-{client|server}-{language}-generator\**
- This directory will contain whatever is needed to build the backend emitter code generator. It will contain whatever folder structure is needed to build that specific native code. It will also contain an isolated ci.yml file which will be the build pipeline for this package.
There is a goal to be able to ship these emitter packages independent from the rest of the packages in this repo as such they by default be excluded from the root pnpm workspace. Any npm package work will be isolated to those directories with a goal of eventually moving to a consistent model so that we can both work in isolation as well as work as a E2E.
For language specific contributing information look for the contributing.md file in that specific lanague emitter folder.