Seems related to the fact that I have not set an avatar yet:

Serde Error:
1 | ...08-25T21:19:44.962Z","image":"","discord":"brwhale","phone":"...
| ^ invalid value: string "", expected relative URL without a base at line 1 column 154

Note that on terminal with fixed width font, that carat points to the string for "image", which is empty.

https://github.com/KittyCAD/kcl-samples/blob/main/.github/workflows/output-from-kcl-samples.yml#L31 runs sed to try to trim the date out and hope for determinism.
Instead, there should be a flag to force the output to be deterministic.

This enable users to refactor kcl code and ensure that the result stays the same. As a project using kcl grows, the ability to safely refactor will become critical in large teams.

This may not always be possible for all formats, and that's useful insight in itself as the tool can assert this explicitly, instead of hoping for the best.

but those work locally if you just run them...

Our homebrew formula downloads a tarball from https://dl.zoo.dev/releases/cli/v0.2.36/kittycad-cli.tar.gz

This tarball is supposed to contain the zoo CLI executables/binaries for various architectures. However, it doesn't. The files aren't binaries, they're XML docs, each of which says

<?xml version='1.0' encoding='UTF-8'?>
<Error>
  <Code>NoSuchKey</Code>
  <Message>The specified key does not exist.</Message>
</Error>

The releases are getting built properly, because https://dl.zoo.dev/releases/cli/v0.2.36/zoo-aarch64-apple-darwin contains actual binaries. So there's something going wrong when we create the tarball for homebrew. The XML error looks like an S3 error, so maybe it's looking for an S3 key called "kittycad" but it's actually called "zoo", or vice-versa.

This has been happening since at least version 0.2.33.

In https://github.com/KittyCAD/format/issues/189 Frank gets a HTTP 400 error from the API, but the CLI doesn't actually show what the error is. It just shows

Unexpected Response: Response { url: Url { scheme: "https", cannot_be_a_base: false, username: "", password: None, host: Some(Domain("api.kittycad.io")), port: None, path: "/file/conversion/ply/gltf", query: None, fragment: None }, status: 400, headers: {"date": "Wed, 13 Sep 2023 18:18:47 GMT", "content-type": "application/json", "content-length": "168", "x-request-id": "9d9d208e-73ed-45cc-988a-6ef30e7e62e9", "strict-transport-security": "max-age=15724800; includeSubDomains", "cf-cache-status": "DYNAMIC", "report-to": "{\"endpoints\":[{\"url\":\"https:\/\/a.nel.cloudflare.com\/report\/v3?s=DuqbgM4pRYX9FL2l3QBWvrUgoYsQWQrcPyB7JtpUyhlAxF%2FI2uK%2BQXpghYFFWvYgxd28rJQZ2zlEV5kL0fpDBLOnjhEAFexYr3DuuaFV%2FPdW4QJQq2lhk5w7QEuE0BoepI8yANg6qOvsQTj%2Bpho%3D\"}],\"group\":\"cf-nel\",\"max_age\":604800}", "nel": "{\"success_fraction\":0,\"report_to\":\"cf-nel\",\"max_age\":604800}", "set-cookie": "__cflb=02DiuForeBv2DVnYSFPtSfNRmR8hoJ4HciYZCksQzHXLG; SameSite=Lax; path=/; expires=Thu, 14-Sep-23 17:18:47 GMT; HttpOnly", "server": "cloudflare", "cf-ray": "80626505b9b0439d-EWR", "alt-svc": "h3=\":443\"; ma=86400"} }

This sucks for a few reasons:

• We're printing raw debug strings, we should print something prettier
• This does not show the HTTP body which contains the actual error message and code

Instead, we should check what the API error is and print them. If there's no API errors, then we should do something else but nicer than this -- not just print raw rust debug strings.

or we should fix the automation

@paultag 's recent modeling-app PR added a KCL linter. It removed the type kcl_lib::errors::KclError 's method get_message_line_column which is being used by the CLI.

Either add the method back or change the CLI to use something different

When trying out the CLI for the first time, I wanted to check if file conversions worked but I don't have some of the software that some of those file types need to open. So I ended up having the AutoDesk Online File Viewer open while I worked.

Since we will have pixel streaming possible from the engine with the work within KittyCAD/engine, I was wondering if it would be worthwhile for our users to have a way to view 3D models, for debugging or otherwise, within the terminal itself without too much effort. I think it fits our retro-with-superpowers brand.

example usages

• kittycad file view suzanne.step: basic viewer, using something like https://github.com/hzeller/timg if the terminal supports it
• kittycad file view --ascii: display the video stream as ASCII characters. Idea from @alteous

The UX of the viewer would be TBD, but maybe as an MVP a sensible default camera position is chosen, and no camera controls are given to the user. Eventually we could support keyboard combinations to pan, zoom, and orbit.

this is for viuer

If you run cargo +1.78 check on this repo, it passes with 0 errors. But if you run cargo +1.79 check it'll fail.

warning: field `description` is never read
--> src/config.rs:49:9
 |
47 | pub struct ConfigOption {
 |            ------------ field in this struct
48 |     pub key: String,
49 |     pub description: String,
 |         ^^^^^^^^^^^
 |
 = note: `#[warn(dead_code)]` on by default

warning: field `tmp_file_override` is never read
--> src/iostreams.rs:38:9
 |
11 | pub struct IoStreams {
 |            --------- field in this struct
...
38 |     pub tmp_file_override: Option<std::fs::File>,
 |         ^^^^^^^^^^^^^^^^^

warning: `zoo` (bin "zoo") generated 2 warnings

Weird. I guess the new rustc is smarter about finding dead code. We have to fix this to make CI pass again.

@greg-kcio encountered a bug when trying to update on Ubuntu. The error is:

No such file or directory (os error 2)

I've attached the strace, but it points to some weird reading of a symlink to the executable itself.

Quickly searching the code, I'm pretty sure it's this line:

// src/cmd_update.rs:30
let current_binary_path = std::env::current_exe()?;

strace.txt

I get the same error on Arch Linux.

https://stripe.com/docs/cli/status

once we have KittyCAD/modeling-app#2521

I run brew tap kittycad/kittycad and then brew install kittycad. I get:

==> Downloading https://formulae.brew.sh/api/formula.jws.json
######################################################################### 100.0%
==> Downloading https://formulae.brew.sh/api/cask.jws.json
######################################################################### 100.0%
==> Fetching kittycad/kittycad/kittycad
==> Downloading https://dl.zoo.dev/releases/cli/v0.2.32/kittycad-cli.tar.gz
Already downloaded: /Users/adamchalmers/Library/Caches/Homebrew/downloads/5452d95c15d036ad4450eee7a2a4a755eb7bc601eb5b4499f97dec9ff5f42559--kittycad-cli.tar.gz
==> Installing kittycad from kittycad/kittycad
Error: An exception occurred within a child process:
  Errno::ENOENT: No such file or directory - aarch64_darwin/zoo

Theory

The homebrew formulae we publish (https://github.com/KittyCAD/homebrew-kittycad/blob/main/Formula/kittycad.rb) references a tar.gz. When I download that tar, it expands into this:

I think our release is creating aarch64_darwin/kittycad instead of aarch64_darwin/zoo.

In the releases there are install instructions for various platforms, but the windows instructions are wrong as they use linux paths e.g.

windows/amd64

# Export the sha256sum for verification.
$ export KITTYCAD_CLI_SHA256="ee4a911b6814262b4e4d83c877654715a6442d401dbd2a104ebf31e82101e024"


# Download and check the sha256sum.
$ curl -fSL "https://dl.kittycad.io/releases/cli/v0.0.9/cli-windows-amd64" -o "/usr/local/bin/kittycad" \
	&& echo "${KITTYCAD_CLI_SHA256}  /usr/local/bin/kittycad" | sha256sum -c - \
	&& chmod a+x "/usr/local/bin/kittycad"


$ echo "kittycad cli installed!"

# Run it!
$ kittycad -h

Changes probably need to be made to cli/basic.mk

Looking at our cli install action might help as it works with a Windows runner, though it still uses bash, perhaps instructions should use powershell. Also adding the cli to %PATH% would be different outside of a github action runner.