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

Re: [PATCH v7] unpack-trees: suggest using 'git stash' when checkout fails

From
Junio C Hamano <gitster@pobox.com>
Date
Mar 12, 2026, 22:40 UTC
Message-ID
<xmqqldfwacyw.fsf@gitster.g>
In-Reply-To
<pull.2233.v7.git.git.1773345901659.gitgitgadget@gmail.com>
"Arsh Srivastava via GitGitGadget" <gitgitgadget@gmail.com> writes:
Show 5 quoted lines
> When a branch switch fails due to local changes and
> new users who are not familiar with the error message often
> get confused about how to move ahead and resolve the issue as
> the previous error message only suggests to commit or stash the changes
> but doesn't explain how to do that or what the next steps are.

The first paragraph is a bit of a run-on and has a misplaced "and"; I cannot quite read and understand this overly long single sentence.

Perhaps the early part can become a bit easier to read with punctuations, and cutting the sentence into two, e.g.,

  When a branch switch fails due to local changes, new users who
  are unfamiliar with the error message often get confused about how
  to move ahead and resolve the issue.

Also it is misleading to say "previous" error message. We talk about the current code in the present tense, to highlight what the problem in the current code is. The _current_ message stops at hinting the commands to be used without giving wordy instructions that are best left to manuals. You may view it as a weakness (which may motivate this patch to be written). But I personally am not so sure that adding words to the existing message would necessarily make it more clear.

> This patch enhances the error message with more specific
> instructions in a concise manner to help users understand
> how to resolve the issue and move their local changes
> safely to the other branch using stash.

As Documentation/SubmittingPatches says, let's instruct the code to "be like so" in imperative mood. E.g., "Enhance the error message..." instead of "This patch enhances...".

By the way, the updated message seems much less concise than the original.

Show 9 quoted lines
>  	msg = advice_enabled(ADVICE_COMMIT_BEFORE_MERGE)
>  	      ? _("Your local changes to the following files would be overwritten by checkout:\n%%s"
> -		  "Please commit your changes or stash them before you switch branches.")
> -	      : _("Your local changes to the following files would be overwritten by checkout:\n%%s");
> +		  "To move you local changes safely to the other branch,\n"
> +		  "Please try 'git stash' followed by 'git checkout <branch>' followed by 'git stash pop' for safe merge."
> +		  )
> +	      : _("Your local changes to the following files would be overwritten by checkout:\n%%s"
> +		  "Please commit your changes or stash them before you switch branches.");

These were already overly long, but the updated one is way too long to be read on end-user's terminal. The source lines are overly long, too.

The original was this:
 	msg = advice_enabled(ADVICE_COMMIT_BEFORE_MERGE)
 	      ? _("Your local changes to the following files would be overwritten by checkout:\n%%s"
		  "Please commit your changes or stash them before you switch branches.")
	      : _("Your local changes to the following files would be overwritten by checkout:\n%%s");
Note that when advice is *NOT* enabled, we only gave
_("Your local changes to the following files would be overwritten by checkout:\n%%s");
without any "advise" in the output.  That is what !advice_enabled() means.
The updated code does this:
	msg = advice_enabled(ADVICE_COMMIT_BEFORE_MERGE)
	      ? _("Your local changes to the following files would be overwritten by checkout:\n%%s"
		  "To move you local changes safely to the other branch,\n"
		  "Please try 'git stash' followed by 'git checkout <branch>' followed by 'git stash pop' for safe merge."
		  )
	      : _("Your local changes to the following files would be overwritten by checkout:\n%%s"
		  "Please commit your changes or stash them before you switch branches.");

to those users who decline the advice, we now show "Please commit...". That is not what !advice_enabled() should trigger, is it?

Also "To move you" -> "To move your".

Also the advice lost the other possiblity of first committing the work in progress on the original branch before switching, yet the new advice message is quite wordy.

Also, using "for safe merge" when the user is performing a "checkout" might be slightly confusing, even if 'stash pop' involves a merge under the hood.

A more concise version might say:
  Try 'git stash && git checkout <branch> && git stash pop' to carry
  your changes to the new branch, or commit your work before switching.

But as I already said, I think the current text may already strike the right balance between being clear and being concise.

Thanks.
Previous: Arsh Srivastava via GitGitGadgetNext: Arsh Srivastava
Message 45 of 62 in “Advice on checkout dirty files”
  1. 0/5 Advice on checkout dirty filesArsh Srivastava via GitGitGadget, Mar 10, 2026
  2. 1/5 diff: handle ANSI escape codes in prefix when calculating diffstat widthLorenzoPegorari via GitGitGadget, Mar 10, 2026
  3. 2/5 t4052: test for diffstat width when prefix contains ANSI escape codesLorenzoPegorari via GitGitGadget, Mar 10, 2026
  4. 3/5 repo: remove unnecessary variable shadowK Jayatheerth via GitGitGadget, Mar 10, 2026
  5. 4/5 The 13th batchJunio C Hamano via GitGitGadget, Mar 10, 2026
  6. 5/5 advice: add stashBeforeCheckout advice for dirty branch switchesArsh Srivastava via GitGitGadget, Mar 10, 2026
  7. Phillip WoodMar 10, 2026
  8. Arsh SrivastavaMar 10, 2026
  9. Arsh SrivastavaMar 10, 2026
  10. Junio C HamanoMar 10, 2026
  11. Arsh SrivastavaMar 10, 2026
  12. Junio C HamanoMar 10, 2026
  13. Arsh SrivastavaMar 10, 2026
  14. Arsh SrivastavaMar 10, 2026
  15. advice: add stashBeforeCheckout advice for dirty branch switchesArsh Srivastava via GitGitGadget, Mar 10, 2026
  16. Patrick SteinhardtMar 10, 2026
  17. Arsh SrivastavaMar 10, 2026
  18. Patrick SteinhardtMar 10, 2026
  19. 0/2 Advice on checkout dirty filesArsh Srivastava via GitGitGadget, Mar 10, 2026
  20. 1/2 advice: add stashBeforeCheckout advice for dirty branch switchesArsh Srivastava via GitGitGadget, Mar 10, 2026
  21. Arsh SrivastavaMar 10, 2026
  22. 2/2 advice: add stashBeforeCheckout advice for dirty branch switches [GSOC]Arsh Srivastava via GitGitGadget, Mar 10, 2026
  23. 0/5 Advice on checkout dirty filesArsh Srivastava via GitGitGadget, Mar 11, 2026
  24. 1/5 advice: add stashBeforeCheckout advice for dirty branch switchesArsh Srivastava via GitGitGadget, Mar 11, 2026
  25. 2/5 advice: add stashBeforeCheckout advice for dirty branch switches [GSOC]Arsh Srivastava via GitGitGadget, Mar 11, 2026
  26. 3/5 unpack-trees: suggesting 'git checkout -m <branch>' with its repercussionsArsh Srivastava via GitGitGadget, Mar 11, 2026
  27. 4/5 Updating tests and unpack-tress.c [GSOC]Arsh Srivastava via GitGitGadget, Mar 11, 2026
  28. 5/5 File updation [GSOC]Arsh Srivastava via GitGitGadget, Mar 11, 2026
  29. Junio C HamanoMar 11, 2026
  30. Arsh SrivastavaMar 11, 2026
  31. 0/3 Advice on checkout dirty filesArsh Srivastava via GitGitGadget, Mar 11, 2026
  32. 1/3 advice: add stashBeforeCheckout advice for dirty branch switchesArsh Srivastava via GitGitGadget, Mar 11, 2026
  33. 2/3 advice: add stashBeforeCheckout advice for dirty branch switches [GSOC]Arsh Srivastava via GitGitGadget, Mar 11, 2026
  34. 3/3 unpack-trees: suggesting 'git checkout -m <branch>' with its repercussionsArsh Srivastava via GitGitGadget, Mar 11, 2026
  35. Junio C HamanoMar 12, 2026
  36. Arsh SrivastavaMar 12, 2026
  37. unpack-trees: suggesting 'git checkout -m' with its repercussionsArsh Srivastava via GitGitGadget, Mar 12, 2026
  38. Junio C HamanoMar 12, 2026
  39. Arsh SrivastavaMar 12, 2026
  40. Junio C HamanoMar 12, 2026
  41. Arsh SrivastavaMar 12, 2026
  42. Junio C HamanoMar 12, 2026
  43. Arsh SrivastavaMar 12, 2026
  44. unpack-trees: suggest using 'git stash' when checkout failsArsh Srivastava via GitGitGadget, Mar 12, 2026
  45. Junio C HamanoMar 12, 2026
  46. Arsh SrivastavaMar 13, 2026
  47. Karthik NayakMar 13, 2026
  48. Arsh SrivastavaMar 13, 2026
  49. Junio C HamanoMar 13, 2026
  50. Arsh SrivastavaMar 13, 2026
  51. Arsh SrivastavaMar 13, 2026
  52. Karthik NayakMar 10, 2026
  53. Arsh SrivastavaMar 10, 2026
  54. Arsh SrivastavaMar 10, 2026
  55. Karthik NayakMar 10, 2026
  56. Arsh SrivastavaMar 10, 2026
  57. Arsh SrivastavaMar 10, 2026
  58. Junio C HamanoMar 10, 2026
  59. Karthik NayakMar 10, 2026
  60. Konstantin RyabitsevMar 14, 2026
  61. Arsh SrivastavaMar 10, 2026
  62. Arsh SrivastavaMar 10, 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.