Hi,
Having the following:
cc_binary(
name = "foo",
srcs = ["foo.cpp"],
data = ["@some//:files"], # Very large, gigabytes
)
Where:
filegroup(
name = "files",
data = [ ... ],
)
And the "@some" repository comes from an "http_archive".
If I run "bazel build //path/to:foo", it downloads GB's of runtime data dependencies, that are not needed to simply build C++ code.
Why are these runtime data dependencies downloaded at "build" time? Is there a way to avoid this, and download only at "run" time? This causes great trouble for build and lint actions that don't depend on the data.
Thanks!