On Mon, Nov 04, 2024 at 01:37:27PM +0100, Benno Evers wrote: > I'm afraid I can't offer much wisdom, but a few thoughts: Thank you. Between this response and a bit of pondering over the last few days, I think I have a firm understanding of the issue and possible paths forward. So here's the series I came up with, which starts by adjusting the tests to be resilient to the later changes, but also to show the existing failure mode. And then the rest of the patches add the performance improvements we've been discussing in the thread. I'll drop the kernel lists from the cc since I think this has gotten well off topic there. [1/4]: t6120: demonstrate weakness in disjoint-root handling [2/4]: t/perf: add tests for git-describe [3/4]: describe: stop digging for max_candidates+1 [4/4]: describe: stop traversing when we run out of names builtin/describe.c | 17 ++++++++++------- t/perf/p6100-describe.sh | 30 ++++++++++++++++++++++++++++++ t/t6120-describe.sh | 15 ++++++++++++--- 3 files changed, 52 insertions(+), 10 deletions(-) create mode 100755 t/perf/p6100-describe.sh -Peff
