Maybe you should write a script that spits out AI description for a commit and then run it for commits without a proper description? Since it doesn’t require any insight from the commit author it should work the same.
Can we at least mention, though, that that’s kind of nonsensical, too? Give me a *very* high-level summary of what changed, but then the rest of the commit message should be the why (unless that’s genuinely obvious, like when adding a feature).
If I actually want to know what changed, I can look at the code changes. I can’t find the why anywhere else, though. Nor can an LLM having to describe those random code changes.
Maybe you should write a script that spits out AI description for a commit and then run it for commits without a proper description? Since it doesn’t require any insight from the commit author it should work the same.
Can we at least mention, though, that that’s kind of nonsensical, too? Give me a *very* high-level summary of what changed, but then the rest of the commit message should be the why (unless that’s genuinely obvious, like when adding a feature).
If I actually want to know what changed, I can look at the code changes. I can’t find the why anywhere else, though. Nor can an LLM having to describe those random code changes.
I’m now tempted to do this for all several thousand commits in the main branch, and at the very least create a better changelog.