
I have something more specific than "generic PR assistant" in mind, so read carefully. What I mean is a model that:
Takes an issue + natural language description as input
Creates a branch
Writes some amount of code relevant to the issue (I will accept something as minimal as "skeleton of the code + tests")
Allows the engineer to review/edit the changes made
Creates and submits a PR
Interacts with CI (not necessarily all CI platforms, but at least one)
Can make additional edits based on CI and reviewer feedback
If it misses one of these steps that's okay.
The model does not need to be open source or even open to the public: if it is announced that Microsoft is using such a thing internally than that can count.
The tool must meet a minimum threshold of "better than not having the tool".
If it's available to the public, then the standard will be "I, vluzko, use this tool (including paying for it if it costs money)".
If it is internal at some org then I will ask people I know at that org to anonymously comment on whether or not they actually prefer the tool.
If I do not know anyone at the org then I will search for credible reports from users that they actually like it. If I can't find any (e.g. the only comments available are from the C-suite and the marketing team) then I will resolve NO, not N/A.
In addition, part of the benefit of the tool must be the code it writes. If the way I/others use it is by immediately deleting all the code it writes (i.e. it's just a nice UI for interacting with the PR cycle) that does not resolve YES.
Additional requirements:
The code generation must be deep learning based.
It does not need to be an LLM.
The other steps do not need to be AI based. If someone writes all the structural code to let a model interact with the PR cycle and then the model just writes a little code, that would resolve YES.
๐ Top traders
# | Name | Total profit |
---|---|---|
1 | แน9 | |
2 | แน1 |
People are also trading
@ElliotDavies A Pull Request (PR) is a process in version control systems like Git where a developer proposes changes to a codebase. Other developers review the changes, discuss modifications, and eventually merge the changes into the main branch when they are approved.
Continuous Integration (CI) is a software development practice where developers integrate their code changes into a shared repository frequently, usually several times a day. Each integration is verified by an automated build and tests to detect errors and conflicts as quickly as possible.