# Development Tips
# RLS/rust-analyzer and IDEs
The most common question on developing Teaclave is how to use Rust IDEs to improve the development experience, e.g., code completions, type hints and cross references. Internally, these features are supported by either RLS or rust-analyzer. Unfortunately, these features are not supported in Teaclave's codebase out-of-box. The reason is that Teaclave has components targeting different environments (SGX enclave and Linux app) which need different set of dependencies (SGX crates and vanilla crates). To support this flexible building and linking process, we are using cmake for our build system. However, there are still ways to workaround and let the analyzer understand the project structures.
When developing SGX enclaves and corresponding dependent crates, you need to
prepare set of cargo files in the root directory to help the analyzer. This
includes Cargo.toml
which can be copied from our build system:
cmake/tomls/Cargo.sgx_trusted_lib.toml
; Cargo.lock
which can be copied from
third_party/crates-sgx/Cargo.lock
; .cargo/config
which can be copied from
third_party/crates-sgx/config
. Similarly, when developing the app parts, you
can copy the cmake/tomls/Cargo.sgx_untrusted_lib.toml
file to the root
directory as Cargo.toml
, and .cargo/config
and Cargo.lock
files can be
found in third_party/crates-io/
directory. Besides, please also change the
last line of directory = "vendor"
to directory = "third_party/crates-sgx/vendor"
or directory ="third_party/crates-io/vendor"
so the dependencies can be resolved automatically. For standalone Rust
applications such as CLI, no Cargo.toml
is needed. After the preparation of
Cargo.toml
in root, RLS/rust-analyzer can understand the projects finally. You
will see type hints and cross references using IDEs with extensions.
NOTE
You can also simply use the script tools/scripts/ide.sh <trusted|untrusted|clean>
to prepare a IDE-friendly developing environment for trusted
part, untrusetd part, or remove the files generated by this script.
# Logging
Teaclave utilizes the env_logger
crate to configure the display of debug logs via environment variables.
Logging is controlled via the TEACLAVE_LOG
environment variables and the value
of this variable is a comma-separated list of logging directives in the
path::to::module=level
form. For example, you can set the environment
TEACLAVE_LOG=attestation=debug
before launching a service to print the debug
level (and higher-level) logs in the attestation
module to stdout/stderr.
There are five logging levels: error
, warn
, info
, debug
and trace
where error represents the highest-priority log level. Furthermore, you can also
filter the results with regular expression by simply put /
followed by a regex
in the directives in the environment variable. You can find more filter usages
in the env_logger
's
document.
NOTE
To prevent sensitive information leakage through logging, for the release build,
we disable all logging (at build time) lower than the info
level. That is,
only error
, warn
and info
logs will be printed.