Hi,
in files where only a single conflict is, I see agents pretty much always forget to remove // weave: run 'weave explain ...' for per-hunk detail, 'weave check' to verify your resolution after fixing the conflict. I think this is because for example in this conflict:
function MyFunction() {
....
....
....
....
....
....
{yieldThirdPartyScriptsEnabled && (
<OptimizationSection
<<<<<<< ours — scope `return (` in function `Performance` (F, confidence: medium)
// refused_by: statement_fold · collision: `title={<T>Optimize Third-Party Scripts</T>}`
title={<T>Optimize third-party scripts</T>}
=======
title={<T>{Dictionary.OptimizeThirdPartyScripts}</T>}
>>>>>>> theirs — scope `return (`
htmlFor="yield-third-party-scripts"
....
....
....
....
....
....
}
// weave: run 'weave explain ...' for per-hunk detail, 'weave check' to verify your resolution
In this case it's just a few lines, but in functions that span hundreds of lines, missing the comment at the end of the file (?) is pretty easy even as a human. I think if there is only a single conflict, it makes sense to put the instructions right in there, e.g.:
>>>>>>> theirs — scope `return (`
htmlFor="yield-third-party-scripts"
// weave: run 'weave explain ...' for per-hunk detail, 'weave check' to verify your resolution
Hi,
in files where only a single conflict is, I see agents pretty much always forget to remove
// weave: run 'weave explain ...' for per-hunk detail, 'weave check' to verify your resolutionafter fixing the conflict. I think this is because for example in this conflict:In this case it's just a few lines, but in functions that span hundreds of lines, missing the comment at the end of the file (?) is pretty easy even as a human. I think if there is only a single conflict, it makes sense to put the instructions right in there, e.g.: