Re: [PATCH v4 5/6] bisect--helper: reimplement `bisect_run` shell function in C

Hi Miriam,

On Tue, 17 Aug 2021, Miriam R. wrote:

El mar, 17 ago 2021 a las 13:42, Johannes Schindelin
([off-list ref]) escribió:

quoted

On Tue, 17 Aug 2021, Miriam Rubio wrote:

quoted

From: Tanushree Tumane <redacted>

[...]
+
+             if (res == 125)
+                     strvec_push(&args, "skip");
+             else if (res > 0)
+                     strvec_push(&args, terms->term_bad);
+             else
+                     strvec_push(&args, terms->term_good);
+
+             res = bisect_state(terms, args.v, args.nr);

Since `args.nr` will always be 1, it would probably be better to use
something like this:

                const char *new_state;

                [...]
                if (res == 125)
                        new_state = "skip";
                else
                        new_state = res > 0 ?
                                terms->term_bad : terms->term_good;

                res = bisect_state(terms, &new_state, 1);

Yes, indeed. I will change it.

quoted

Also: I think at this stage, an equivalent to `cat
"$GIT_DIR/BISECT_RUN"` is missing.

In the previous patch series (v3), I implemented the equivalent to the
cat command but I understood reviewers wanted to print the output to the
user, so I reverted my changes for this version.
https://lore.kernel.org/git/20210411095538.34129-4-mirucam@gmail.com/ (local)

I am a bit confused: doesn't `bisect_state()` write to the `BISECT_RUN`
file? If so, I think we do need to show the contents by opening the file
and piping it to `stdout`.

FWIW I read
https://lore.kernel.org/git/CAP8UFD3X24F3qgefHpi00PM-KUk+vcqxwy2Dbngbyj7ciavCVQ@mail.gmail.com/ (local)
to mean the same thing, although I have to admit that I am not 100%
certain.

Just to make sure: with this patch, at the end of a `git bisect` run, the
user is shown the commit message of the first bad commit?

Ciao,
Dscho