git/list[1] front-page[2] threads[3] people[4] search[5] about
wed 2026-10-07 17:28 UTC

Re: first bisection step takes quite a while

From
UKUwe Kleine-König <u.kleine-koenig@baylibre.com>
Date
Feb 21, 2025, 09:28 UTC
Message-ID
<4hx5uvjy7mzntb5zp6o4dg3ut44i46bthyfuera3lnbpbcvrey@kbo3ejype7ae>
In-Reply-To
<xmqqa5agcbx6.fsf@gitster.g>
Hello Junio,
On Thu, Feb 20, 2025 at 05:40:53PM -0800, Junio C Hamano wrote:
> Comments?

It's long time ago that I looked into the git source code and I guess many things have changed since then.

Anyhow, here comes my thought about how finding a bisection point could work.

Pick the middle commit of `git rev-list --topo-order $bad ^$allgood`. Lets assume this are 10000 commits. Check the weight of commit[5000]. Depending on how much the weight is off from 5000 make a bigger or a smaller step up or down to find the next commit to check. So a scaled bisection on the topo-order commit list. I think that doesn't necessarily finds a best bisection point, but I havn't thought about that a lot.

Best regards Uwe

Previous: Christian CouderNext: Junio C Hamano
Message 6 of 12 in “first bisection step takes quite a while”
  1. Uwe Kleine-KönigFeb 20, 2025
  2. D. Ben KnobleFeb 20, 2025
  3. Junio C HamanoFeb 20, 2025
  4. Junio C HamanoFeb 21, 2025
  5. Christian CouderFeb 21, 2025
  6. Uwe Kleine-KönigFeb 21, 2025
  7. Junio C HamanoFeb 21, 2025
  8. Junio C HamanoFeb 21, 2025
  9. Christian CouderFeb 21, 2025
  10. Christian CouderFeb 21, 2025
  11. Junio C HamanoFeb 22, 2025
  12. D. Ben KnobleFeb 24, 2025

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.