(cpp) Add C++ query implementation - #201
Conversation
…aseStartedOrderBy
…CaseStarted, FindAllTestRunHookFinished
…Steps, findAllTestStepStarted, findAllUndefinedParameterTypes, findAttachmentsBy
… FindTestCaseStartedBy
…edBy, FindTestRunHookStartedBy, FindTestRunStarted
…lization in tests
- Change CompilationDatabase path in .clangd to current directory - Add compile_commands.json to .gitignore - Create custom target for clangd compile commands in CMakeLists.txt - Update find_package for nlohmann_json and cucumber_messages
…nd host-conan presets
| @@ -0,0 +1,722 @@ | |||
| # The MIT License (MIT) | |||
| # | |||
| # Copyright (c) 2024 JFrog | |||
| # against. For releases use a tag matching the version required in | ||
| # cpp/CMakeLists.txt (find_package) and cpp/conanfile.py; for cross-repo | ||
| # feature work, set this to the messages branch under test. | ||
| MESSAGES_REF: 8cd6cacecd0a657095c93b069c850f51009e0499 |
There was a problem hiding this comment.
This should be an exact version, updateable by Renovate, and preferably not kept here.
mpkorstanje
left a comment
There was a problem hiding this comment.
LGTM except for the way messages fetched. I assume Conan will fix that?
|
@mpkorstanje I am going to revert the conan changes in all honesty. I've been trying to get it to work in a satisfying and useable way. But its usability isn't on-par as to what cucumber needs sadly. All the packages that we depend on are available in system packages, so I am updating the test builds to use those available to linux, macos and windows. The downside is we don't have control over the exact installed packages, OTOH, if the tests work with an older version than that should be fine too. Yes conan can work with github releases but setting it up properly is a tremendous amount of manual updating dependencies. I'll have a look at the comment w.r.t. the hash a bit later after I removed the conan things. Currently very busy on giving the cucumber/pretty-formatter a cpp makeover. And doing so I am updating Query as well to update some useability issues. p.s. I thought I created this as a draft, but apparently vscode tricked me in to creating a non-draft. Will downgrade to draft. Sorry about that. |
|
Okay, no problem.
I'd prefer it if we did depend on a specific version. Waiting for system packages is a bit too slow too make a comfortable feedback cycle. But we should depend on a specific version, not a hash. and Renovate should be able to update that. We've done that before so I hope we can reuse that mechanism. |
🏷️ What kind of change is this?
📋 Checklist: