# [PATCH 2/2] Bisect: implement "bisect dunno" to mark untestable revisions.

4 messages from 2007-10-08 to 2007-10-08. Participants: Christian Couder, Johannes Schindelin.
Thread: https://gitlist.dev/t/10187

## Christian Couder, 2007-10-08 03:34

Subject: [PATCH 2/2] Bisect: implement "bisect dunno" to mark untestable revisions.
Message-ID: <20071008053450.a52d7c5e.chriscool@tuxfamily.org>
URL: https://gitlist.dev/e/20071008053450.a52d7c5e.chriscool%40tuxfamily.org

```
When there are some dunno revisions, we add the '--bisect-all'
option to "git rev-list --bisect-vars". Then we filter out the
dunno revisions from the result of the rev-list command, and we
modify the "bisect_rev" var accordingly.

We don't always use "--bisect-all" because it is slower
than "--bisect-vars" or "--bisect".

When we cannot find for sure the first bad commit because of
dunno commits, we print the hash of each possible first bad
commit and then we exit with code 2.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
---
 git-bisect.sh               |  127 +++++++++++++++++++++++++++++++++++++++++--
 t/t6030-bisect-porcelain.sh |   71 ++++++++++++++++++++++++
 2 files changed, 193 insertions(+), 5 deletions(-)

diff --git a/git-bisect.sh b/git-bisect.sh
index 388887a..c556318 100755
--- a/git-bisect.sh
+++ b/git-bisect.sh
@@ -17,6 +17,8 @@ git bisect replay <logfile>
         replay bisection log.
 git bisect log
         show bisect log.
+git bisect dunno [<rev>...]
+        mark <rev>... untestable revisions.
 git bisect run <cmd>...
         use <cmd>... to automatically bisect.'
 
@@ -143,7 +145,7 @@ bisect_write_bad() {
 
 bisect_good() {
 	bisect_autostart
-        case "$#" in
+	case "$#" in
 	0)    revs=$(git rev-parse --verify HEAD) || exit ;;
 	*)    revs=$(git rev-parse --revs-only --no-flags "$@") &&
 		test '' != "$revs" || die "Bad rev input: $@" ;;
@@ -153,7 +155,6 @@ bisect_good() {
 		rev=$(git rev-parse --verify "$rev^{commit}") || exit
 		bisect_write_good "$rev"
 		echo "git-bisect good $rev" >>"$GIT_DIR/BISECT_LOG"
-
 	done
 	bisect_auto_next
 }
@@ -164,6 +165,28 @@ bisect_write_good() {
 	echo "# good: "$(git show-branch $rev) >>"$GIT_DIR/BISECT_LOG"
 }
 
+bisect_dunno() {
+	bisect_autostart
+	case "$#" in
+	0)    revs=$(git rev-parse --verify HEAD) || exit ;;
+	*)    revs=$(git rev-parse --revs-only --no-flags "$@") &&
+		test '' != "$revs" || die "Bad rev input: $@" ;;
+	esac
+	for rev in $revs
+	do
+		rev=$(git rev-parse --verify "$rev^{commit}") || exit
+		bisect_write_dunno "$rev"
+		echo "git-bisect dunno $rev" >>"$GIT_DIR/BISECT_LOG"
+	done
+	bisect_auto_next
+}
+
+bisect_write_dunno() {
+	rev="$1"
+	echo "$rev" >"$GIT_DIR/refs/bisect/dunno-$rev"
+	echo "# dunno: "$(git show-branch $rev) >>"$GIT_DIR/BISECT_LOG"
+}
+
 bisect_next_check() {
 	missing_good= missing_bad=
 	git show-ref -q --verify refs/bisect/bad || missing_bad=t
@@ -206,17 +229,104 @@ bisect_auto_next() {
 	bisect_next_check && bisect_next || :
 }
 
+search_dunno() {
+	_hash="$1"
+	_dunno="$2"
+
+	for _val in $_dunno ; do
+		case $_hash in $_val) return 1 ;; esac
+	done
+	return 0
+}
+
+filter_dunno() {
+	_eval="$1"
+	_dunno="$2"
+
+	if [ -z "$_dunno" ]; then
+		eval $_eval
+		return
+	fi
+
+	# Let's parse the output of:
+	# "git rev-list --bisect-vars --bisect-all ..."
+	eval $_eval | while read hash line
+	do
+		case "$VARS,$FOUND,$TRIED,$hash" in
+			# We display some vars.
+			1,*,*,*) echo "$hash $line" ;;
+
+			# Split line.
+			,*,*,---*) ;;
+
+			# We had nothing to search.
+			,,,bisect_rev*)
+				echo "bisect_rev="
+				VARS=1
+				;;
+
+			# We did not find a good bisect rev.
+			# This should happen only if the "bad"
+			# commit is also a "dunno" commit.
+			,,*,bisect_rev*)
+				echo "bisect_rev=$TRIED"
+				VARS=1
+				;;
+
+			# We are searching.
+			,,*,*)
+				TRIED="${TRIED:+$TRIED|}$hash"
+				if search_dunno "$hash" "$_dunno" ; then
+					echo "bisect_rev=$hash"
+					echo "bisect_tried=\"$TRIED\""
+					FOUND=1
+				fi
+				;;
+
+			# We have already found a rev to be tested.
+			,1,*,bisect_rev*) VARS=1 ;;
+			,1,*,*) ;;
+
+			# ???
+			*) die "filter_dunno error " \
+			    "VARS: '$VARS' " \
+			    "FOUND: '$FOUND' " \
+			    "TRIED: '$TRIED' " \
+			    "hash: '$hash' " \
+			    "line: '$line'"
+			;;
+		esac
+	done
+}
+
+exit_if_dunno_commits () {
+	_tried=$1
+	if expr "$_tried" : ".*[|].*" > /dev/null ; then
+		echo "There are only 'dunno' commit left to test."
+		echo "The first bad commit could be any of:"
+		echo "$_tried" | sed -e 's/[|]/\n/g'
+		echo "We cannot bisect more!"
+		exit 2
+	fi
+}
+
 bisect_next() {
-        case "$#" in 0) ;; *) usage ;; esac
+	case "$#" in 0) ;; *) usage ;; esac
 	bisect_autostart
 	bisect_next_check good
 
+	dunno=$(git for-each-ref --format='%(objectname)' \
+		"refs/bisect/dunno-*" | tr '[\012]' ' ') || exit
+
+	BISECT_OPT=''
+	test -n "$dunno" && BISECT_OPT='--bisect-all'
+
 	bad=$(git rev-parse --verify refs/bisect/bad) &&
 	good=$(git for-each-ref --format='^%(objectname)' \
 		"refs/bisect/good-*" | tr '[\012]' ' ') &&
-	eval="git rev-list --bisect-vars $good $bad --" &&
+	eval="git rev-list --bisect-vars $BISECT_OPT $good $bad --" &&
 	eval="$eval $(cat "$GIT_DIR/BISECT_NAMES")" &&
-	eval=$(eval "$eval") &&
+	eval=$(filter_dunno "$eval" "$dunno") &&
 	eval "$eval" || exit
 
 	if [ -z "$bisect_rev" ]; then
@@ -224,11 +334,16 @@ bisect_next() {
 		exit 1
 	fi
 	if [ "$bisect_rev" = "$bad" ]; then
+		exit_if_dunno_commits "$bisect_tried"
 		echo "$bisect_rev is first bad commit"
 		git diff-tree --pretty $bisect_rev
 		exit 0
 	fi
 
+	# We should exit here only if the "bad"
+	# commit is also a "dunno" commit (see above).
+	exit_if_dunno_commits "$bisect_rev"
+
 	echo "Bisecting: $bisect_nr revisions left to test after this"
 	echo "$bisect_rev" >"$GIT_DIR/refs/heads/new-bisect"
 	git checkout -q new-bisect || exit
@@ -363,6 +478,8 @@ case "$#" in
         bisect_bad "$@" ;;
     good)
         bisect_good "$@" ;;
+    dunno)
+        bisect_dunno "$@" ;;
     next)
         # Not sure we want "next" at the UI level anymore.
         bisect_next "$@" ;;
diff --git a/t/t6030-bisect-porcelain.sh b/t/t6030-bisect-porcelain.sh
index 03cdba5..7f41a46 100755
--- a/t/t6030-bisect-porcelain.sh
+++ b/t/t6030-bisect-porcelain.sh
@@ -71,6 +71,63 @@ test_expect_success 'bisect start with one bad and good' '
 	git bisect next
 '
 
+# $HASH1 is good, $HASH4 is bad, we dunno about $HASH3
+# but $HASH2 is bad,
+# so we should find $HASH2 as the first bad commit
+test_expect_success 'bisect dunno: successfull result' '
+	git bisect reset &&
+	git bisect start $HASH4 $HASH1 &&
+	git bisect dunno &&
+	git bisect bad > my_bisect_log.txt &&
+	grep "$HASH2 is first bad commit" my_bisect_log.txt &&
+	git bisect reset
+'
+
+# $HASH1 is good, $HASH4 is bad, we dunno about $HASH3 and $HASH2
+# so we should not be able to tell the first bad commit
+# among $HASH2, $HASH3 and $HASH4
+test_expect_success 'bisect dunno: cannot tell between 3 commits' '
+	git bisect start $HASH4 $HASH1 &&
+	git bisect dunno || return 1
+
+	if git bisect dunno > my_bisect_log.txt
+	then
+		echo Oops, should have failed.
+		false
+	else
+		test $? -eq 2 &&
+		grep "first bad commit could be any of" my_bisect_log.txt &&
+		! grep $HASH1 my_bisect_log.txt &&
+		grep $HASH2 my_bisect_log.txt &&
+		grep $HASH3 my_bisect_log.txt &&
+		grep $HASH4 my_bisect_log.txt &&
+		git bisect reset
+	fi
+'
+
+# $HASH1 is good, $HASH4 is bad, we dunno about $HASH3
+# but $HASH2 is good,
+# so we should not be able to tell the first bad commit
+# among $HASH3 and $HASH4
+test_expect_success 'bisect dunno: cannot tell between 2 commits' '
+	git bisect start $HASH4 $HASH1 &&
+	git bisect dunno || return 1
+
+	if git bisect good > my_bisect_log.txt
+	then
+		echo Oops, should have failed.
+		false
+	else
+		test $? -eq 2 &&
+		grep "first bad commit could be any of" my_bisect_log.txt &&
+		! grep $HASH1 my_bisect_log.txt &&
+		! grep $HASH2 my_bisect_log.txt &&
+		grep $HASH3 my_bisect_log.txt &&
+		grep $HASH4 my_bisect_log.txt &&
+		git bisect reset
+	fi
+'
+
 # We want to automatically find the commit that
 # introduced "Another" into hello.
 test_expect_success \
@@ -99,6 +156,20 @@ test_expect_success \
      grep "$HASH4 is first bad commit" my_bisect_log.txt &&
      git bisect reset'
 
+# $HASH1 is good, $HASH5 is bad, we dunno about $HASH3
+# but $HASH4 is good,
+# so we should find $HASH5 as the first bad commit
+HASH5=
+test_expect_success 'bisect dunno: add line and then a new test' '
+	add_line_into_file "5: Another new line." hello &&
+	HASH5=$(git rev-parse --verify HEAD) &&
+	git bisect start $HASH5 $HASH1 &&
+	git bisect dunno &&
+	git bisect good > my_bisect_log.txt &&
+	grep "$HASH5 is first bad commit" my_bisect_log.txt &&
+	git bisect reset
+'
+
 #
 #
 test_done
-- 
1.5.3.4.208.g996ad

```

## Johannes Schindelin, 2007-10-08 03:49

Subject: Re: [PATCH 2/2] Bisect: implement "bisect dunno" to mark untestable revisions.
Message-ID: <Pine.LNX.4.64.0710080444290.4174@racer.site>
URL: https://gitlist.dev/e/Pine.LNX.4.64.0710080444290.4174%40racer.site
In-Reply-To: <20071008053450.a52d7c5e.chriscool@tuxfamily.org>

```
Hi,

On Mon, 8 Oct 2007, Christian Couder wrote:

> diff --git a/git-bisect.sh b/git-bisect.sh
> index 388887a..c556318 100755
> --- a/git-bisect.sh
> +++ b/git-bisect.sh
> @@ -143,7 +145,7 @@ bisect_write_bad() {
>  
>  bisect_good() {
>  	bisect_autostart
> -        case "$#" in
> +	case "$#" in

White space breakage.

> @@ -153,7 +155,6 @@ bisect_good() {
>  		rev=$(git rev-parse --verify "$rev^{commit}") || exit
>  		bisect_write_good "$rev"
>  		echo "git-bisect good $rev" >>"$GIT_DIR/BISECT_LOG"
> -

?

> @@ -164,6 +165,28 @@ bisect_write_good() {
>  	echo "# good: "$(git show-branch $rev) >>"$GIT_DIR/BISECT_LOG"
>  }
>  
> +bisect_dunno() {
> +	bisect_autostart
> +	case "$#" in
> +	0)    revs=$(git rev-parse --verify HEAD) || exit ;;
> +	*)    revs=$(git rev-parse --revs-only --no-flags "$@") &&
> +		test '' != "$revs" || die "Bad rev input: $@" ;;
> +	esac
> +	for rev in $revs
> +	do
> +		rev=$(git rev-parse --verify "$rev^{commit}") || exit
> +		bisect_write_dunno "$rev"
> +		echo "git-bisect dunno $rev" >>"$GIT_DIR/BISECT_LOG"

Should the last line not be put into bisect_write_dunno?  OTOH this is the 
only call site of that function, so I strongly doubt that the function 
(consisting of 3 lines, where the first is 'rev="$1"') is necessary at 
all.

> @@ -206,17 +229,104 @@ bisect_auto_next() {
>  	bisect_next_check && bisect_next || :
>  }
>  
> +search_dunno() {
> +	_hash="$1"
> +	_dunno="$2"
> +
> +	for _val in $_dunno ; do
> +		case $_hash in $_val) return 1 ;; esac
> +	done

This would be faster as

	case " $1" in " $2") return 1 ;; esac

I guess.

But as I said in the other reply, I think this logic belongs into the C 
core, instead of generating mostly useless information, passing it down to 
the script, and filtering it out again.

Thanks,
Dscho

```

## Christian Couder, 2007-10-08 05:34

Subject: Re: [PATCH 2/2] Bisect: implement "bisect dunno" to mark untestable revisions.
Message-ID: <200710080734.23878.chriscool@tuxfamily.org>
URL: https://gitlist.dev/e/200710080734.23878.chriscool%40tuxfamily.org
In-Reply-To: <Pine.LNX.4.64.0710080444290.4174@racer.site>

```
Le lundi 8 octobre 2007, Johannes Schindelin a écrit :
> Hi,
>
> On Mon, 8 Oct 2007, Christian Couder wrote:
> > diff --git a/git-bisect.sh b/git-bisect.sh
> > index 388887a..c556318 100755
> > --- a/git-bisect.sh
> > +++ b/git-bisect.sh
> > @@ -143,7 +145,7 @@ bisect_write_bad() {
> >
> >  bisect_good() {
> >  	bisect_autostart
> > -        case "$#" in
> > +	case "$#" in
>
> White space breakage.

The patch tries to fix some white space breakages.

> > @@ -153,7 +155,6 @@ bisect_good() {
> >  		rev=$(git rev-parse --verify "$rev^{commit}") || exit
> >  		bisect_write_good "$rev"
> >  		echo "git-bisect good $rev" >>"$GIT_DIR/BISECT_LOG"
> > -
>
> ?

It also removes this unneeded blank line.

> > @@ -164,6 +165,28 @@ bisect_write_good() {
> >  	echo "# good: "$(git show-branch $rev) >>"$GIT_DIR/BISECT_LOG"
> >  }
> >
> > +bisect_dunno() {
> > +	bisect_autostart
> > +	case "$#" in
> > +	0)    revs=$(git rev-parse --verify HEAD) || exit ;;
> > +	*)    revs=$(git rev-parse --revs-only --no-flags "$@") &&
> > +		test '' != "$revs" || die "Bad rev input: $@" ;;
> > +	esac
> > +	for rev in $revs
> > +	do
> > +		rev=$(git rev-parse --verify "$rev^{commit}") || exit
> > +		bisect_write_dunno "$rev"
> > +		echo "git-bisect dunno $rev" >>"$GIT_DIR/BISECT_LOG"
>
> Should the last line not be put into bisect_write_dunno?  OTOH this is
> the only call site of that function, so I strongly doubt that the
> function (consisting of 3 lines, where the first is 'rev="$1"') is
> necessary at all.

Well, there are "bisect_write_bad" and "bisect_write_good" that already do 
the same thing as "bisect_write_dunno". In fact I thought that it was 
better to just copy "bisect_dunno" from "bisect_good" 
and "bisect_write_dunno" from "bisect_write_good".

If needed I can send another patch to factorise these functions.

> > @@ -206,17 +229,104 @@ bisect_auto_next() {
> >  	bisect_next_check && bisect_next || :
> >  }
> >
> > +search_dunno() {
> > +	_hash="$1"
> > +	_dunno="$2"
> > +
> > +	for _val in $_dunno ; do
> > +		case $_hash in $_val) return 1 ;; esac
> > +	done
>
> This would be faster as
>
> 	case " $1" in " $2") return 1 ;; esac
>
> I guess.

I will try your suggestion and send an updated patch. Thanks.

> But as I said in the other reply, I think this logic belongs into the C
> core, instead of generating mostly useless information, passing it down
> to the script, and filtering it out again.

Yeah, it's not efficient.

Best regards,
Christian.

```

## Johannes Schindelin, 2007-10-08 05:36

Subject: Re: [PATCH 2/2] Bisect: implement "bisect dunno" to mark untestable revisions.
Message-ID: <Pine.LNX.4.64.0710080632510.4174@racer.site>
URL: https://gitlist.dev/e/Pine.LNX.4.64.0710080632510.4174%40racer.site
In-Reply-To: <200710080734.23878.chriscool@tuxfamily.org>

```
Hi,

On Mon, 8 Oct 2007, Christian Couder wrote:

> Le lundi 8 octobre 2007, Johannes Schindelin a ?crit :
>
> > On Mon, 8 Oct 2007, Christian Couder wrote:
> > > diff --git a/git-bisect.sh b/git-bisect.sh
> > > index 388887a..c556318 100755
> > > --- a/git-bisect.sh
> > > +++ b/git-bisect.sh
> > > @@ -143,7 +145,7 @@ bisect_write_bad() {
> > >
> > >  bisect_good() {
> > >  	bisect_autostart
> > > -        case "$#" in
> > > +	case "$#" in
> >
> > White space breakage.
> 
> The patch tries to fix some white space breakages.
> 
> > > @@ -153,7 +155,6 @@ bisect_good() {
> > >  		rev=$(git rev-parse --verify "$rev^{commit}") || exit
> > >  		bisect_write_good "$rev"
> > >  		echo "git-bisect good $rev" >>"$GIT_DIR/BISECT_LOG"
> > > -
> >
> > ?
> 
> It also removes this unneeded blank line.

Both laudable changes; alas, they distracted me.

> > > @@ -164,6 +165,28 @@ bisect_write_good() {
> > >  	echo "# good: "$(git show-branch $rev) >>"$GIT_DIR/BISECT_LOG"
> > >  }
> > >
> > > +bisect_dunno() {
> > > +	bisect_autostart
> > > +	case "$#" in
> > > +	0)    revs=$(git rev-parse --verify HEAD) || exit ;;
> > > +	*)    revs=$(git rev-parse --revs-only --no-flags "$@") &&
> > > +		test '' != "$revs" || die "Bad rev input: $@" ;;
> > > +	esac
> > > +	for rev in $revs
> > > +	do
> > > +		rev=$(git rev-parse --verify "$rev^{commit}") || exit
> > > +		bisect_write_dunno "$rev"
> > > +		echo "git-bisect dunno $rev" >>"$GIT_DIR/BISECT_LOG"
> >
> > Should the last line not be put into bisect_write_dunno?  OTOH this is
> > the only call site of that function, so I strongly doubt that the
> > function (consisting of 3 lines, where the first is 'rev="$1"') is
> > necessary at all.
> 
> Well, there are "bisect_write_bad" and "bisect_write_good" that already 
> do the same thing as "bisect_write_dunno". In fact I thought that it was 
> better to just copy "bisect_dunno" from "bisect_good" and 
> "bisect_write_dunno" from "bisect_write_good".

If they also are called by just one site, and also do not do the complete 
printing to the log in the function (but also in the caller), I think they 
are not really worth it, either.

> If needed I can send another patch to factorise these functions.

That's not up to me to decide.  I'm just saying what I dislike.

Please do not take my criticism as a sign of a personal attack; if I did 
not find your patch worthwhile, I would not bother to respond.  So in a 
way, it is my way to show my appreciation for your work that I review and 
criticize it; for efficiency, I do not mention what I like ;-)

Ciao,
Dscho

```
