Dear Git Organization, I'm writing to follow up on my Google Summer of Code proposal, "Move Existing Tests to a Unit Testing Framework." After careful consideration and based on our discussions, I've refined the project summary and deliverables to focus specifically on Move existing tests to a unit testing framework.This email aims to gather feedback and initiate a discussion on the feasibility and implementation of this idea. --- Google Summer of Code 2024 Name: Aishwarya N University: SASTRA University Program: B.Tech Electrical and Electronics Engineering Organization: Git Project Idea: Move existing tests to a unit testing framework GitHub: https://github.com/Aishwarya-Narayanan03 Email: aishnana.03@xxxxxxxxx Timezone: IST (GMT+5:30) About Me My name is Aishwarya N., and I am soon to graduate from SASTRA Deemed University in Thanjavur, India, with a B.Tech. in Electrical and Electronics Engineering. With graduation only two months away, I'm ready to plunge into the exciting world of open source and offer my expertise to a project that enables developers. The power of building something from nothing through programming is a truly special and empowering feeling. It's no wonder it resonates with me! The Git project particularly appeals to me. Contributing to the tool that supports innumerable software developments provides an opportunity to have a genuine, global impact. Why Me? Adaptable and Open to Challenges: I thrive in lively surroundings and see new challenges as opportunities to learn and progress. I acknowledge that mistakes are unavoidable on the route to mastery, and I am not afraid to learn from them. An avid learner: My thirst for knowledge is boundless. I'm always looking for new methods to improve myself and push the boundaries of technology. Collaborative Spirit: While I'm quite satisfied working independently, I believe that collaborating is crucial for accomplishing ambitious goals. I am an adept communicator who is eager to work with the brilliant Git community. Why Git? The fast-paced world of software piques my curiosity, and Git, a basic tool for collaboration and version control, is a necessary skill set for success in this area of expertise. But, beyond its functionality, Git is fascinating.I'm excited to contribute to the project that underpins numerous developer workflows. Project Summary The Git project has recently implemented a new unit testing framework to improve code quality and maintainability. Currently, numerous legacy test cases are stored in separate test helper files (t/helper/test-*.c) written in C, as well as test scripts (t/*.sh) created in the shell to call these helpers. This project focuses on migrating these existing tests to a unified format using the new framework within dedicated C source files (t/unit-tests/t-*.c).This conversion will offer various advantages: By unifying tests in the new framework, the codebase structure becomes more consistent. Improved test readability and maintainability: The new framework makes it easier to write clear and maintainable tests. By utilising the new framework, we can expand our testing capabilities and features beyond the current approach. Before GSoC Microproject:Avoiding suppressing git's exit code in test scripts Learning Git Codebase 1. Understanding Test Scripts: Exploring the test scripts in the Git project's t- directory provided personal experience with real-world Shell scripting used for integration testing. This helpd me understand how Git's features are tested and confirmed. 2.Reading Commits: The reference to commit c6f44e1da5 demonstrated how to handle Git commands in test scripts. Analysing this commit allowed me to examine the adjustments done to address the issue of suppressing Git's exit code, as well as get insight into the codebase's structure and convention. 3.Navigating the Git Repository: Searching for instances of problematic code and finding solutions within the Git repository enhanced my ability to successfully navigate and search a big codebase. Best Practices 1. Avoiding Pitfalls:Avoiding typical mistakes in test scripts, such as suppressing exit codes, can lead to more dependable testing infrastructure. 2.Understanding Pipe Usage: Knowing when and how to use pipes in Shell scripting is critical. This microproject describes circumstances in which pipes may mistakenly suppress exit codes and provides suggestions for proper use. Connecting With Mentors: 1.Engaging with Git contributors: This microproject allowed me to connect with mentors and Git project contributors. Seeking clarity on specific code practices or debating changes to test scripts encourages involvement in the open-source community. 2.Receiving comments: Submitting updates or queries about the microproject helped in receving useful comments from experienced developers.This feedback helped me learn the Git codebase and improve my scripting skills. 3.Community Involvement: Participating in discussions about the microproject on platforms such as Git email lists or forums allowed me to communicate with a diverse community of developers while sharing information and experiences. Related Works https://lore.kernel.org/git/CAHCXyj3U69qyhYewOLY9hN2rvi_5ZuSxQEBJbDxrKefm9MzVWg@xxxxxxxxxxxxxx/ - Microproject Approach https://lore.kernel.org/git/ZgQffea0krKmZUEt@tanuki/ - [RFC PATCH] Description: While printf might be unlikely to fail here,I felt the change aligns with the project's goal of ensuring Git command exit codes are captured. This approach is more robust and avoids potential issues in the future. Status: Declined https://public-inbox.org/git/xmqqttkqwfwe.fsf@gitster.g/ - [RFC PATCH] Fix Git command exit code suppression in test script t2104-update-index-skip-worktree.sh Description: This patch increases the robustness of Git's testing framework by guaranteeing that Git commands' exit codes are appropriately examined during test execution. Previously, suppressed exit codes allowed tests to pass despite Git command failures. This fix stores the output of Git commands in variables and examines their exit codes to detect errors.The discussion focuses on improvements to the patch submission process, such as adhering to coding principles and referencing appropriate documentation for proper formatting and test script adjustments. Status: Declined https://public-inbox.org/git/CAHCXyj1hUVNNuCOgsNv4GJUi79_o9iWZDvV8Ocz3DodreYoL7g@xxxxxxxxxxxxxx/ - GSoC 2024 [PATCH v2] Fix Git command exit code suppression in test script t2104-update-index-skip-worktree.sh Description: This patch resolves an issue in Git test scripts in which the exit code of git ls-files -t may be disregarded, resulting in tests passing despite Git command failures. The new version guarantees that Git commands used in pipelines produce captured output and perform adequate exit code checks. It also enhances code style by adhering to established rules. This adds to more robust Git testing by ensuring the identification and reporting of Git command errors within test scripts. Status: Under Review Porting Unit Tests: [1]https://lore.kernel.org/git/cover.1692297001.git.steadmon@xxxxxxxxxx/ This patch series provides unit test functionality for the Git project. The series includes the following patches: A project plan document that outlines the goals for introducing unit tests, as well as a review of potential frameworks and the features used to assess them. An implementation of the TAP unit test framework with a sample unit test and Makefile integration. Changes were made to the Git build system so that the unit tests could be run in CI. The TAP framework was selected because of the following advantages: Simple to use and comprehend. Widely used and integrates seamlessly with existing Git tools. The project plan document mentions a number of outstanding TODOs, including how to ensure tests run on various OS platforms and how to gather unit test statistics. This is a well-written patch series with a straightforward and logical approach. The documentation is comprehensive and instructive, and the code is well-organized and simple to grasp. Here are some extra factors to consider: The patch series discusses collecting unit test statistics. We could clarify on the type of statistics we want to collect? How do we intend to deal with flaky tests? [2]https://lore.kernel.org/git/20240112102743.1440-1-ach.lumap@xxxxxxxxx/ This is an email thread about a patch series submitted by Achu Luma to port the unit tests for C character classification functions from the legacy helper tool to the new unit testing framework ([Outreachy][PATCH v5] Port helper/test-ctype.c to unit-tests/t-ctype.c). Here are the major changes implemented in the patch series: The unit tests for C character classification routines (isdigit(), isspace(), etc.) are moved from the historical approach, which used the test-tool command test-tool ctype in t/helper/test-ctype.c, to the new unit testing framework (t/unit-tests/test-lib.h). Refactors the tests to use the framework's testing macros (TEST() and check_*()). Added tests for handling EOF (end-of-file). Overall, the patch series was well received and thought to be a positive development. The reviewers made various suggestions for improvement, such as expanding on the types of statistics the author is interested in gathering for the unit tests and how to deal with flaky tests. [3] https://lore.kernel.org/git/20240112102122.1422-1-ach.lumap@xxxxxxxxx/ This patch migrates unit tests for the advise_if_enabled functionality from the legacy approach to a new unit testing framework. Here's a breakdown of the changes: Legacy approach: Used the test-tool advise command in t/helper/test-advise.c. Relied on helper functions from test-tool.h. Tests were written in shell script (t/t0018-advice.sh). New approach: Uses the new unit testing framework with TEST() macros (t/unit-tests/t-advise.c). Includes necessary headers for advise.h, config.h, and setup.h. Tests are written in C code. The patch removes the following files from the legacy approach: t/helper/test-advise.c t/t0018-advice.sh The new unit test file t/unit-tests/t-advise.c includes three tests covering different scenarios: Advice is printed when the configuration variable is unset. Advice is printed when the configuration variable is set to true. Advice is not printed when the configuration variable is set to false. Overall, this patch improves the codebase by using a more standardized and maintainable approach for unit testing. [4]https://lore.kernel.org/git/20240205162506.1835-1-ach.lumap@xxxxxxxxx/ Refactoring 64-bit Pre-requisite Checks in date.h This is a patch for the Linux kernel which refactors pre-requisite checks for 64-bit systems into reusable functions in the date.h header file. Context: The codebase recently added a unit testing framework. Some shell tests use a helper tool that checks for 64-bit prerequisites (TIME_IS_64BIT and TIME_T_IS_64BIT). These checks are also needed in the new unit tests. To avoid duplication, the patch moves these checks to date.h as inline functions. Changes: Two new inline functions are added to date.h: check_prereq_TIME_IS_64BIT: Checks if the size of timestamp_t is 8 bytes (indicating a 64-bit system). check_prereq_TIME_T_IS_64BIT: Checks if the size of time_t is 8 bytes (indicating a 64-bit system). The shell test functions in t/helper/test-date.c are modified to use the new functions instead of directly checking the sizes. Benefits: Reduces code duplication. Improves code maintainability. Simplifies porting shell test code to the new unit test framework. Next Steps: A following commit will port the relevant code from t/helper/test-date.c to the new unit test file t/unit-tests/t-date.c, using the newly introduced functions for pre-requisite checks. [5]Unit Testing in Git: https://github.com/git/git/blob/master/Documentation/technical/unit-tests.txt discusses unit testing for the Git project. This gives an idea of the background and importance of unit testing in the project applying to. During GSoc The main goals of this project are: 1. Understand the existing "reftable" unit tests: Examine the present implementation of the "reftable" unit tests in 't0032-reftable-unittest.sh' to ensure that you understand their functionality and purpose. 2. Learn the new unit testing framework: Get a thorough grasp of Git's new unit testing framework, including setup, usage, and recommended practices. 3.Ensuring test coverage and reliability: Check that the converted tests have sufficient code coverage and accurately simulate the expected behaviour of the "reftable" capability. Test and validate the new tests thoroughly to ensure their reliability. 4. Documentation and code comments: Document the conversion process, including any issues encountered and the reasoning behind design decisions. Update the code comments and documentation to reflect the changes to the tests. Deliverables ü All legacy test cases previously residing in t/helper/test-*.c and t/*.sh files will be converted to the new unit testing framework within dedicated C source files in t/unit-tests/t-*.c. ü The migrated tests will be thoroughly executed and validated to ensure complete coverage of all existing test functionalities. ü Documentation explaining the migration procedure, as well as revisions to existing documentation. ü The final report summarises the project's accomplishments, problems, and future recommendations. Expected timeline Community bonding period (May 1 - May 26) Familiarise myself with the Git project's code and contribution standards. Discuss the project plan and milestones with mentors. Establish the development environment and tools required for the project. Phase 1: Analysis and Planning (May 27 - June 15) Examine the current reftable unit tests and the unit testing framework documentation. Plan the migration process by identifying the test cases and capabilities that will be covered. Break down tasks and establish a precise project timeline. Phase 2: Transition to C-based Unit Tests (June 16–July 15) Begin rebuilding the existing shell scripts as C-based unit tests. Ensure that the migration process is accurate and comprehensive, while retaining test coverage. Communicate with mentors on a regular basis to receive feedback and guidance. Phase 3: Testing and Validation (July 16–August 5) Evaluate the migrated unit tests against different scenarios and edge situations. Debug and resolve any errors or anomalies that arise during testing. Before finalising, ensure that the unit tests are proper and reliable. Phase 4: Documentation and Reporting (August 6–August 26) Document the migration process, including the obstacles encountered and solutions implemented. Update the relevant documentation to reflect the modifications to the reftable unit tests. Prepare a detailed report summarising the project's results and contributions. Availability My university holidays begin on June 30. However, considering this is my final year with only three classes and major project, I believe it will not be a problem. I intent to work 6 hours a day, 5 days a week this summer.I have no scheduled trips except for GIM-SRIP from May 6th to 31st. I will inform Mentors advance notice of any changes in availability. After GSoC I realise that it would be much more beneficial if our GSoC participants could stay after the event. And that is exactly what I want to do. I think doing open-source, especially with a community that fuels a popular development tool, is simply fascinating. And, in addition to being refreshing, such a community allows me to learn and grow, with the possibility of making more significant contributions if I continue to participate.While this is my first foray into open-source development and GSoC, I am excited by the prospect of contributing to the ongoing success of Git. I have been actively learning and gaining a strong foundation in Git commands, programming languages, and version control concepts. Closing Remarks Finally, I'd like to express my gratitude to the community and, in particular, my mentors, Patrick Steinhardt,Christian Couder,Kaartic Sivaraam,Karthik Nayak,Junio C Hamano. They did and continue to do an exceptional task of maintaining and empowering the Git open-source community, as well as offering much-needed and kind assistance to a new contributor like me. --- Sincerely, Aishwarya Narayanan
