Filter by reset
- Everywhere
166 - Issues
46 - Pull Requests
40 - Comments
80
Implementing a duration and the relative percentage for each command in a test seems like a great idea. We can use our existing block syntax and just overlook it during the test: ––– duration: 0.123ms (10.2%) –––
enhancement
…"git operation failed": The button becomes invisible when I hover over it: Too much empty space here which makes the prompt's font too small: "your command or question" is confusing. It makes users wonder things like: "Is there a list of supported …
Improve CLT output with user command
We should improve our CLI action and provide users with a given command that they can run to validate the test locally. In the same output, we should dump information about the script and real command execution.
CI: Display CLT Test EDIT link with a given domain as a parameter in action
Now with our UI, we can not just post error tests to comments in pull requests but also provide a link to edit it in the UI. The link looks like this: {UI_HOST}/?test_path=core%2Ftest-auto-schema.rec&branch=master&failed_tests[]=core%2Ftest-auto-schema. …
Currently, we keep Rust binaries in a repository that is not very efficient. We should think about introducing a CI pipeline that will build it, and as per CLT tool usage, we can just automatically download it when required to the local system and keep …
Closer integration with GitHub, so all changes are saved right away in a GitHub repo Support for Docker checkpoints Show progress while a test is running Syntax highlighting Add Docker image name as a comment (like metadata) Editable code blocks …
We should think about the easiest way to integrate what we researched into the UI that is still in the branch. We should add a CLI binary that will utilize our MCP server and will be able to produce tests with user prompts through the UI and also allow us …
From: https://gitlab.com/manticoresearch/clt/-/issues/5 While doing a scenario to test mysqldump with combined tasks no mcl, mcl only, mcl id with json, MVA, text fields, we found that it's a good idea to have the possibility to support dynamic variables …
Bug: CLT crashes on test written with AI while it looks like correct
There is crash on test generated AI. We should investigate and fix it. Also, we should improve the MCP server to be error-prone, and in case of a crash or wrong exit codes, it should also return an error. Comprehensive test of all full-text search …
Reusable blocks in the interactive mode
From: https://gitlab.com/manticoresearch/clt/-/issues/10 We now have support for reusable blocks, but we should also incorporate this feature into interactive mode. The task involves: Contemplating possible implementation methods. As proposed, we could …
Remove escape sequences from outputs
We should remove all bash sequences from generated output before we write it to the rep file to make it easier and more natural to compare outputs with different IDEs that may not render them or simply do not display them.
est::size_S
Skip adding succeed tests to summary
We should add to summary only those tests that failed same we do in comments
Description: HTTP requests sent through Buddy without ; echo at the end cause the request to hang indefinitely, and the test does not complete. The query works correctly when executed via MySQL CLI (mysql -h0 -P9306 -e "show version;"). The issue occurs …
bug
UI to continously compose and run test
We should implement a simple UI that will perform the following functions: Expose an interface allowing users to enter commands for execution and display two rows of compared output - one modifiable and another from replay Automatically rerun tests when …
enhancement
CLT improvements and bug fixes
We should add some improvements to CLT and investigate bugs Currently we cannot see OK in outputs by default, looks like at some point in time it stopped working We should add possibility to disable single line matching as requested by @PavelShilin89
CLT UI fails to process %{NUMBER}
To make things easier and try to make AI follow proper structure and better blend flow, there is an idea to develop an MCP server and adapt it with the flow to create tests from scratch.
Extra pipe for all output from commands that we run
We should consider looking into the possibility of adding an extra pipe (e.g., > &3) that would allow us to store launched commands' output and review or retrieve it at the end of the test for debugging purposes. Also, as previously considered, it would …
Let's add support for ---comment--- totally ignored by CLT when replaying the test. It will make it much easier to understand what the author means when testing something.
Make CLT to run from global PATH
Currently when we add path to CLT in PATH we are still unable to use it due to failed to find required files to run in specified path