feat(cli): add cdktf watch command

[ansgarm]

Resolves #801

The initial implementation of the new cdktf watch command also contains an
experiment in regards to separating the underlying invocation of a synth and of
Terraform for init, plan and apply from the presentation layer (which is in our
case a CLI build with React components using ink). The other commands (e.g.
cdktf deploy) which we currently have, make use of React Hooks for
orchestrating the underlying steps that are required. Quite early it turned out
to be unreasonable to implement the dynamic nature of the watch command using
those same tools. Instead we opted for trying out a separate process and using a
GraphQL api to connect the two. The main reason for this (as opposed to e.g.
ProtoBuf or REST) was the superior tooling support that exists for this in
TypeScript and React. However, with this prototype now in the open, we will
carefully examine it and assess the pros and cons before jumping ship.

[DanielMSchmidt]

That's a really cool new feature 🙌

[DanielMSchmidt]

• When I run this I see starting server and connecting messages. If I didn't read through the implementation I'd be confused to what server is starting and what it's connecting to.
• I see errors twice currently

- UI wise these should be in separate lines I beli

eve - It does not show the last elements being deployed

- Terraform assets seem to be not included in the watch, I changed sth in my application I deploy via NullResource and include via TerraformAsset but no update was triggered

[ansgarm]

I see errors twice currently

@DanielMSchmidt Have you set the CDKTF_LOG_LEVEL to something other than the default?

[DanielMSchmidt]

@ansgarm I don't think so, I believe it was unset. But shouldn't the logs go only into the cdktf.log file anyways?

[ansgarm]

@ansgarm I don't think so, I believe it was unset. But shouldn't the logs go only into the cdktf.log file anyways?

@DanielMSchmidt If you set the log level but not the DISABLE_LOGGING flag, it will print the errors afaik

[DanielMSchmidt]

@DanielMSchmidt If you set the log level but not the DISABLE_LOGGING flag, it will print the errors afaik

Yeah that could have been the case :)

[ansgarm]

@DanielMSchmidt
Fixed multiple resources that were displayed side by side ✅

[ansgarm]

Also massive spoiler for the Office Hours later on 😉

[ansgarm]

Terraform assets seem to be not included in the watch, I changed sth in my application I deploy via NullResource and include via TerraformAsset but no update was triggered

@DanielMSchmidt Is that file you changed by chance in your .gitignore?

Theoretically it should watch every file that is not in your .gitignore for synthesizing and it should watch all files in the working directory of your target stack for deploying.

[DanielMSchmidt]

@DanielMSchmidt Is that file you changed by chance in your .gitignore?

No I referenced the docker backend with TerraformAsset that is in ../../application/backend and changed the index.js in there (in the docker e2e example)

[github-actions]

I'm going to lock this pull request because it has been closed for 30 days. This helps our maintainers find and focus on the active issues. If you've found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.