git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Ejecting things more aggressively out of 'seen'?

From
Patrick Steinhardt <ps@pks.im>
Date
Nov 3, 2026, 07:30 UTC
Message-ID
<aumNkxzE7wfa519p@pks.im>
In-Reply-To
<xmqqv76fjjhg.fsf@gitster.g>
On Mon, Nov 02, 2026 at 08:17:31AM -0800, Junio C Hamano wrote:
Show 10 quoted lines
> Patrick Steinhardt <ps@pks.im> writes:
> 
> > One state that confuses newcomers quite often is "Will merge to 'next'?"
> > with a question mark at the end. If we were to document these we shuld
> > probably also document that state.
> 
> Hmph, really?  Just like the subject of this e-mail message, a
> trailing question mark is my way of saying "I am planning to do
> this, but if you think it is a bad idea, please stop me with an
> explanation why I shouldn't".

The question that newcomers often have here is whether they need to do anything about this. They basically wonder whether that question is directed at them specifically. Maybe rewording it to "Will merge to next [on <day>] unless told otherwise" may help clarify?

Show 22 quoted lines
> >>  - A topic goes into the "Waiting for response" state once reviewed,
> >>    where it remains until the author responds.  After staying in
> >>    this state for a month or so, it is demoted to the "Stalled"
> >>    state, and eventually dropped.
> >> 
> >> We should probably have a rule that says "a topic that did not get
> >> reviewed for N weeks is ejected from 'seen'".
> >
> > I think that's reasonable. Becoming a bit more aggressive there doesn't
> > make too much of a difference anyway -- if a topic gets evicted from
> > "seen" and then later on regains interest then it can still be included
> > into "seen" again.
> >
> > By gut feeling I'd use the same rule as for "Waiting for response".
> 
> An obvious alternative is to queue only topics for which I can, by
> judging reviews on the list, at least say "Will merge to 'next'?"
> with enough confidence.  It would reduce the size of 'seen' and the
> friction felt by contributors when their topic is once queued on
> 'seen' and later ejected unceremoniously through no fault of their
> own, other than their topic being forgettable or uninspiring to
> others.

Wouldn't we lose the ability to integrate topics with one another early on to see how they interact? And it makes me wonder whether this will overall slow down progress of a patch series if it gets merged to "seen" at a later point.

Show 6 quoted lines
> I do not know which one is better: educating contributors that their
> topics being uninspiring to others is something they should not feel
> bad about and giving our reviewers easier access to these candidate
> topics, or being strict and mechanical and queuing only topics that
> some human reviewers with a certain level of reputation have looked
> at and have not just shown interest in, but have said are ready.

Overall I would say "strict and mechanical" is a good property to have, because it's also easier to explain. But I don't think we should put the bar too high for integrating into "seen" so that we still retain the ability to test the topic together with all the other in-flight topics well ahead of being merged to "next".

So maybe instead of someone saying that a patch series is ready, the bar should be that the patch series has gotten at least one human review that was somewhat favorable on the overall direction?

> I can do either.  It is less emotional stress for me to do the
> latter.  Between hitting 'seen' and then later being discarded, and
> not getting any response or reaction at all, I am not sure which one
> our contributors find more acceptable.

Hard to say. In any case, whatever we choose, it shouldn't only be acceptable for newcomers but also make _your_ life as easy as reasonably possible.

Patrick
Previous: Junio C HamanoNext: Junio C Hamano
Message 13 of 25 in “What's cooking in git.git (Jul 2026, #12)”
  1. Junio C HamanoJul 27, 2026
  2. Phillip WoodJul 29, 2026
  3. Junio C HamanoJul 29, 2026
  4. Phillip WoodJul 29, 2026
  5. Junio C HamanoJul 29, 2026
  6. Phillip WoodAug 5, 2026
  7. Junio C HamanoAug 5, 2026
  8. Ejecting things more aggressively out of 'seen'?Junio C Hamano, Nov 2, 2026
  9. Patrick SteinhardtNov 2, 2026
  10. Junio C HamanoNov 2, 2026
  11. Phillip WoodNov 2, 2026
  12. Junio C HamanoNov 6, 2026
  13. Patrick SteinhardtNov 3, 2026
  14. Junio C HamanoNov 3, 2026
  15. Junio C HamanoNov 3, 2026
  16. Matt HunterJul 31, 2026
  17. Junio C HamanoJul 31, 2026
  18. Harald NordgrenJul 30, 2026
  19. Matt HunterJul 31, 2026
  20. Phillip WoodAug 3, 2026
  21. Junio C HamanoAug 3, 2026
  22. Phillip WoodAug 4, 2026
  23. Phillip WoodAug 3, 2026
  24. Phillip WoodJul 29, 2026
  25. Junio C HamanoJul 29, 2026

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.