build

Description of the problem / feature request:

glob throws an evaluation error when it detects an infinite recursion in a
directory traversal. However, it detects this recursion even if the specific
subpath leading to that recursion is excluded via the exclude=[] argument.

This means that recursive symlinks cannot be ignored in glob expressions,
even if they occur in subtrees that

The local files produced by SAVE ARTIFACT ... AS LOCAL ... always give the same modification time, even when the file content changed. It leads to Earthly targets depended on local artifacts incorrectly cached.

To demonstrate:

FROM busybox

date:
    RUN --no-cache date | tee date
    RUN date -r ./date
    SAVE ARTIFACT date AS LOCAL .

print-date:
    COPY ./date da

PropertyFunctionMSBuildDivide test uses "6.5536" string to represent an expected value. I have Czech regional settings and we are using comma as the decimal symbol, so this test is failing. Local fix is easy - represent the value as number and invoke ToString(). Or we can make the TestPropertyFunction method generic and call ToString() on generic expected parameter.

We are using font-awesome V4 and should migrate to font-awesome V5 which comes with its own vue.js module: https://github.com/FortAwesome/vue-fontawesome

Migration includes removing old font-awesome V4 module and changing all existing icons to new vue.js tag.

This is blocked until #114 is merged which comes with first initial integration.

Some of the repos that we import has ".go" in the file path. eg. https://github.com/nats-io/nats.go.

When using packr v2 to build packr parses the folder path as a file and reports error ".../nats.go" is a directory.

The issue is in findAllGoFiles in jam/parser/finder.go

I am happy to raise a PR if you like

Thanks to using PyO3 (pantsbuild/pants#13526), error handling is now much more ergonomic. Notably, we no longer need to hold the GIL to create a PyErr.

It'd be great to chip away at how much we call .unwrap() in our code, which should make our code more reliable. Instead, probably the only place unwrapping should happen is in our #[pyfunction] and #[pymethods], i

What version of Garble and Go are you using?

$ garble version
v0.4.1-0.20210929123233-e7320ec9c034
$ go version
go version go1.17.2 linux/amd64

What environment are you running Garble on?

Dune systematically buffers the output of commands so that their output is not mixed up during parallel builds. However, this is not ideal for interactive commands that require direct access to the terminal.

Currently, the only workaround is to run dune with -j 1 --no-buffer which is not ideal since it completely disables parallelism and also require a specific dune invocation.

Another ide