Fix cbor encoding for uint64_t's #442
Conversation
There was a problem hiding this comment.
Hello @Ian-Flury, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!
Summary of Changes
Hello! Gemini here, ready to provide a summary of this pull request to help everyone get up to speed quickly. This PR, authored by Ian-Flury, aims to fix an issue in the CBOR encoding logic where unsigned integers, specifically uint64_t, were being incorrectly emitted as signed integers. The changes involve adjusting the Writer class to handle unsigned types separately and adding a new test case to verify the correct encoding of both signed and unsigned 64-bit integers according to the CBOR specification (RFC 8949).
Highlights
- Fix CBOR Unsigned Integer Encoding: The core issue addressed is the incorrect encoding of unsigned integers (like
uint64_t) as signed integers in the CBOR writer. This PR introduces a specific check for unsigned types to ensure they are encoded using the appropriate CBOR major type for unsigned integers. - Add Test for Integer Signedness: A new test file and test case have been added to specifically verify that both signed (
int64_t) and unsigned (uint64_t) integers are encoded correctly according to the CBOR specification, using the distinct major types for each.
Changelog
- include/rfl/cbor/Writer.hpp
- Added a new
if constexprcondition to specifically check forstd::is_unsignedbefore the generalstd::is_integralcheck. - Ensured that unsigned types are cast to
std::uint64_tand encoded usingencoder_->uint64_value.
- Added a new
- tests/cbor/test_integers.cpp
- Added a new test file
test_integers.cpp. - Added a new test case
test_integers_signednessto verify the correct CBOR encoding bytes for largeuint64_tandint64_tvalues. - Included expected byte sequences based on RFC 8949 for unsigned (0x1B) and negative (0x20) integer major types.
- Used
memcmpto compare the generated CBOR output against the expected byte sequences.
- Added a new test file
Using Gemini Code Assist
The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.
Invoking Gemini
You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.
| Feature | Command | Description |
|---|---|---|
| Code Review | /gemini review |
Performs a code review for the current pull request in its current state. |
| Pull Request Summary | /gemini summary |
Provides a summary of the current pull request in its current state. |
| Comment | @gemini-code-assist | Responds in comments when explicitly tagged, both in issue comments and review comments. |
| Help | /gemini help |
Displays a list of available commands. |
Customization
To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.
Limitations & Feedback
Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.
You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.
Unsigned bits, a tricky plight,
Encoded wrong, not shining bright.
A fix applied,
Now verified,
CBOR bytes are finally right.
Footnotes
-
Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩
There was a problem hiding this comment.
Code Review
This PR correctly addresses the CBOR encoding issue for unsigned 64-bit integers by adding a specific handling path for std::is_unsigned types. The change in Writer.hpp is clear and effective. The new test case in test_integers.cpp is a valuable addition and thoroughly checks the signedness behavior with a well-chosen boundary value (0xffffffffffffffff). The comments explaining the CBOR byte structure in the test are particularly helpful.
I have a couple of suggestions regarding the byte buffer comparison method in the tests to improve its robustness.
Summary of Findings
- Test comparison robustness: The
memcmpusage in the new tests (test_integers.cpplines 27 and 36) could lead to undefined behavior (if the actual buffer is shorter than expected) or false positives (if the actual buffer is longer but has a matching prefix). It's recommended to use directstd::vectorcomparison withEXPECT_EQafter appropriate type conversion (e.g., tostd::vector<unsigned char>) for safer and more idiomatic GTest assertions. This ensures both size and content are fully verified.
Merge Readiness
The core fix for CBOR encoding in Writer.hpp is well-implemented and directly addresses the described issue. The accompanying tests are a great addition. However, there's a high-severity issue identified in the new test's comparison logic that could affect test reliability by potentially causing crashes or missing discrepancies. I recommend addressing this test issue to ensure the validation is robust before merging. As a reviewer, I am not authorized to approve pull requests, so please ensure further necessary reviews and approvals are obtained.
Signed-off-by: David Corbeil <8996584+dcorbeil@users.noreply.github.com> Signed-off-by: Ian Flury <ianflury123@gmail.com>
58537e0 to
1fe1700
Compare
|
@Ian-Flury , thanks for your contribution! |
A fix for cbor encoding; unsigned integers were being emitted as signed ints.