Jonathan Ellis

We can have multiple CMs, though. Having a single executor shared by all of them seems like the right thing architecturally.

Maybe that doesn't matter in the brave new virtual thread world. We could try out CF-owns-FragmentExecutor.

when we have multiple fragments corresponding to a given file, we can build a composite of all the fragments, separated by ellipsis; sort both old and new by the order...

> @jbellis has your change request been addressed? thanks! sorry for the slow reply. updated concerns: # Boilerplate 1. createAnalyzer(project) should become a default method that forwards to abstract createAnalyzer(project,...

I did some cleanup and merged manually. (Going forward, pls push to origin so we can collaboratively push to the branch if necessary.)

yes, we're hardcoding "ignore .brokk" but we should respect .gitignore and also Brokk's "exclude from CI" list (do we need a separate ci-exclude list or can we switch that to...

on restart it showed as expected possibly relevant: restart has 3d648348654e650881a1bb10644d55db3af304ff, previous did not

in the meantime, explicit rebuild (Context -> Refresh Code Intelligence) is a workaround

1. parseOwnerRepoFlexible and validateFullRepoName are slightly divergent. The former rejects raw slugs containing double slashes, but validateFullRepoName uses Splitter.omitEmptyStrings(), which could accept "owner//repo" after collapsing empty segments. 2. Extract the...

- GitRepo::applyGitHubAuthentication should be package-private; code that invokes it should live in GitRepoRemote - #collaborating toggling between renderers and row heights is done in several places, can we extract a...