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

Re: [PATCH 09/29] bisect: libify `check_good_are_ancestors_of_bad` and its dependents

From
Johannes Schindelin <johannes.schindelin@gmx.de>
Date
Jan 20, 2020, 22:20 UTC
Message-ID
<nycvar.QRO.7.76.6.2001202313130.46@tvgsbejvaqbjf.bet>
In-Reply-To
<20200120143800.900-10-mirucam@gmail.com>
Hi,
On Mon, 20 Jan 2020, Miriam Rubio wrote:
Show 60 quoted lines
> From: Pranit Bauva <pranit.bauva@gmail.com>
>
> Since we want to get rid of git-bisect.sh it would be necessary to
> convert those exit() calls to return statements so that errors can be
> reported.
>
> Emulate try catch in C by converting `exit(<positive-value>)` to
> `return <negative-value>`. Follow POSIX conventions to return
> <negative-value> to indicate error.
>
> Turn `exit()` to `return` calls in `check_good_are_ancestors_of_bad()`.
>
> Code that turns -11(early success code) to 0 from
> `check_good_are_ancestors_of_bad()` has been moved to
> `cmd_bisect__helper()`.
>
> Handle the return value in dependent function `bisect_next_all()`.
>
> Mentored-by: Christian Couder <chriscool@tuxfamily.org>
> Mentored by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
> Signed-off-by: Pranit Bauva <pranit.bauva@gmail.com>
> Signed-off-by: Tanushree Tumane <tanushreetumane@gmail.com>
> Signed-off-by: Miriam Rubio <mirucam@gmail.com>
> ---
>  bisect.c                 | 42 ++++++++++++++++++++++++++--------------
>  builtin/bisect--helper.c | 12 ++++++++++--
>  2 files changed, 37 insertions(+), 17 deletions(-)
>
> diff --git a/bisect.c b/bisect.c
> index 367258b0dd..2b80597a1d 100644
> --- a/bisect.c
> +++ b/bisect.c
> @@ -865,9 +865,10 @@ static int check_ancestors(struct repository *r, int rev_nr,
>   *
>   * If that's not the case, we need to check the merge bases.
>   * If a merge base must be tested by the user, its source code will be
> - * checked out to be tested by the user and we will exit.
> + * checked out to be tested by the user and we will return.
>   */
> -static void check_good_are_ancestors_of_bad(struct repository *r,
> +
> +static int check_good_are_ancestors_of_bad(struct repository *r,
>  					    const char *prefix,
>  					    int no_checkout)
>  {
> @@ -876,8 +877,15 @@ static void check_good_are_ancestors_of_bad(struct repository *r,
>  	int fd, rev_nr, res = 0;
>  	struct commit **rev;
>
> -	if (!current_bad_oid)
> -		die(_("a %s revision is needed"), term_bad);
> +	/*
> +	 * We don't want to clean the bisection state
> +	 * as we need to get back to where we started
> +	 * by using `git bisect reset`.
> +	 */
> +	if (!current_bad_oid) {
> +		res = error(_("a %s revision is needed"), term_bad);
> +		goto done;
> +	}

Why not just return here? Ah, there is a `filename` that was allocated... it is too bad that we have a mailing-list based review, as the hunk context simply cannot be extended in a mail.

Personally, I think it would be nicer to split the definition of `filename` from its declaration and move it _after_ this conditional code, so that we can `return` right away.

However, there is a more pressing issue than that: `die()` exits with exit code 128, so in keeping with the idea to hand down negative exit codes as return values, should we not assign `res = -128` here?

Show 20 quoted lines
>
>  	/* Check if file BISECT_ANCESTORS_OK exists. */
>  	if (!stat(filename, &st) && S_ISREG(st.st_mode))
> @@ -893,18 +901,20 @@ static void check_good_are_ancestors_of_bad(struct repository *r,
>  	if (check_ancestors(r, rev_nr, rev, prefix))
>  		res = check_merge_bases(rev_nr, rev, no_checkout);
>  	free(rev);
> -	if(res)
> -		exit(res == -11 ? 0 : -res);
> -
> -	/* Create file BISECT_ANCESTORS_OK. */
> -	fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
> -	if (fd < 0)
> -		warning_errno(_("could not create file '%s'"),
> -			      filename);
> -	else
> -		close(fd);
> +
> +	if (!res)
> +	{

We usually put the `{` on the same line as the `if` condition (like you did in the `if (!current_bad_oid)` line above.

The rest looks reasonable. Thank you, Johannes

Show 57 quoted lines
> +		/* Create file BISECT_ANCESTORS_OK. */
> +		fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);
> +		if (fd < 0)
> +			warning_errno(_("could not create file '%s'"),
> +				      filename);
> +		else
> +			close(fd);
> +	}
>   done:
>  	free(filename);
> +	return res;
>  }
>
>  /*
> @@ -975,7 +985,9 @@ int bisect_next_all(struct repository *r, const char *prefix, int no_checkout)
>  	if (read_bisect_refs())
>  		die(_("reading bisect refs failed"));
>
> -	check_good_are_ancestors_of_bad(r, prefix, no_checkout);
> +	res = check_good_are_ancestors_of_bad(r, prefix, no_checkout);
> +	if (res)
> +		return res;
>
>  	bisect_rev_setup(r, &revs, prefix, "%s", "^%s", 1);
>  	revs.limited = 1;
> diff --git a/builtin/bisect--helper.c b/builtin/bisect--helper.c
> index 826fcba2ed..5e0f759d50 100644
> --- a/builtin/bisect--helper.c
> +++ b/builtin/bisect--helper.c
> @@ -666,7 +666,8 @@ int cmd_bisect__helper(int argc, const char **argv, const char *prefix)
>
>  	switch (cmdmode) {
>  	case NEXT_ALL:
> -		return bisect_next_all(the_repository, prefix, no_checkout);
> +		res = bisect_next_all(the_repository, prefix, no_checkout);
> +		break;
>  	case WRITE_TERMS:
>  		if (argc != 2)
>  			return error(_("--write-terms requires two arguments"));
> @@ -713,5 +714,12 @@ int cmd_bisect__helper(int argc, const char **argv, const char *prefix)
>  		return error("BUG: unknown subcommand '%d'", cmdmode);
>  	}
>  	free_terms(&terms);
> -	return !!res;
> +	/*
> +	 * Handle early success
> +	 * From check_merge_bases > check_good_are_ancestors_of_bad > bisect_next_all
> +	 */
> +	if (res == -11)
> +		res = 0;
> +
> +	return res < 0 ? -res : res;
>  }
> --
> 2.21.1 (Apple Git-122.3)
>
>
Previous: Miriam RubioNext: Christian Couder
Message 11 of 54 in “Finish converting git bisect to C part 1”
  1. Miriam RubioJan 20, 2020
  2. 03/29 bisect: use the standard 'if (!var)' way to check for 0Miriam Rubio, Jan 20, 2020
  3. 01/29 bisect--helper: convert `vocab_*` char pointers to char arraysMiriam Rubio, Jan 20, 2020
  4. 02/29 bisect--helper: change `retval` to `res`Miriam Rubio, Jan 20, 2020
  5. 07/29 bisect: libify `bisect_checkout`Miriam Rubio, Jan 20, 2020
  6. 08/29 bisect: libify `check_merge_bases` and its dependentsMiriam Rubio, Jan 20, 2020
  7. Johannes SchindelinJan 20, 2020
  8. Miriam R.Jan 21, 2020
  9. 04/29 run-command: make `exists_in_PATH()` non-staticMiriam Rubio, Jan 20, 2020
  10. 09/29 bisect: libify `check_good_are_ancestors_of_bad` and its dependentsMiriam Rubio, Jan 20, 2020
  11. Johannes SchindelinJan 20, 2020
  12. Christian CouderJan 21, 2020
  13. Miriam R.Jan 21, 2020
  14. 05/29 bisect--helper: introduce new `decide_next()` functionMiriam Rubio, Jan 20, 2020
  15. 10/29 bisect: libify `handle_bad_merge_base` and its dependentsMiriam Rubio, Jan 20, 2020
  16. Johannes SchindelinJan 20, 2020
  17. Christian CouderJan 21, 2020
  18. Miriam R.Jan 21, 2020
  19. 11/29 bisect: libify `bisect_next_all`Miriam Rubio, Jan 20, 2020
  20. Johannes SchindelinJan 20, 2020
  21. Christian CouderJan 21, 2020
  22. Johannes SchindelinJan 30, 2020
  23. 12/29 bisect--helper: reimplement `bisect_next` and `bisect_auto_next` shell functions in CMiriam Rubio, Jan 20, 2020
  24. Johannes SchindelinJan 30, 2020
  25. Miriam R.Jan 31, 2020
  26. Christian CouderFeb 17, 2020
  27. Johannes SchindelinFeb 17, 2020
  28. 13/29 bisect--helper: finish porting `bisect_start()` to CMiriam Rubio, Jan 20, 2020
  29. 06/29 bisect: libify `exit_if_skipped_commits` to `error_if_skipped*` and its dependentsMiriam Rubio, Jan 20, 2020
  30. Johannes SchindelinJan 20, 2020
  31. Christian CouderJan 21, 2020
  32. Miriam R.Jan 21, 2020
  33. 14/29 bisect--helper: retire `--bisect-clean-state` subcommandMiriam Rubio, Jan 20, 2020
  34. 15/29 bisect--helper: retire `--next-all` subcommandMiriam Rubio, Jan 20, 2020
  35. 17/29 bisect--helper: reimplement `bisect_state` & `bisect_head` shell functions in CMiriam Rubio, Jan 20, 2020
  36. 16/29 bisect--helper: reimplement `bisect_autostart` shell function in CMiriam Rubio, Jan 20, 2020
  37. 19/29 bisect--helper: retire `--write-terms` subcommandMiriam Rubio, Jan 20, 2020
  38. 18/29 bisect--helper: retire `--check-expected-revs` subcommandMiriam Rubio, Jan 20, 2020
  39. 20/29 bisect--helper: reimplement `bisect_log` shell function in CMiriam Rubio, Jan 20, 2020
  40. 21/29 bisect--helper: reimplement `bisect_replay` shell function in CMiriam Rubio, Jan 20, 2020
  41. 22/29 bisect--helper: retire `--bisect-write` subcommandMiriam Rubio, Jan 20, 2020
  42. 24/29 bisect--helper: retire `--bisect-autostart` subcommandMiriam Rubio, Jan 20, 2020
  43. 25/29 bisect--helper: retire `--bisect-auto-next` subcommandMiriam Rubio, Jan 20, 2020
  44. 23/29 bisect--helper: use `res` instead of return in BISECT_RESET case optionMiriam Rubio, Jan 20, 2020
  45. 26/29 bisect--helper: reimplement `bisect_skip` shell function in CMiriam Rubio, Jan 20, 2020
  46. 27/29 bisect--helper: retire `--check-and-set-terms` subcommandMiriam Rubio, Jan 20, 2020
  47. 28/29 bisect--helper: reimplement `bisect_visualize()`shell function in CMiriam Rubio, Jan 20, 2020
  48. 29/29 bisect--helper: reimplement `bisect_run` shell function in CMiriam Rubio, Jan 20, 2020
  49. Johannes SchindelinJan 20, 2020
  50. Christian CouderJan 20, 2020
  51. Johannes SchindelinJan 30, 2020
  52. Junio C HamanoJan 30, 2020
  53. Miriam R.Jan 21, 2020
  54. Johannes SchindelinJan 30, 2020

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.