From 8ce79af03f04b59f1dc2b2cada5230255211d963 Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 11:18:54 +0000
Subject: [PATCH 1/7] Adding draft tool script for assessing the risk of
 changes to the MPS.

---
 tool/git-riskiness | 293 +++++++++++++++++++++++++++++++++++++++++++++
 tool/git-risks     |  37 ++++++
 2 files changed, 330 insertions(+)
 create mode 100644 tool/git-riskiness
 create mode 100755 tool/git-risks

diff --git a/tool/git-riskiness b/tool/git-riskiness
new file mode 100644
index 0000000000..8ab6950af2
--- /dev/null
+++ b/tool/git-riskiness
@@ -0,0 +1,293 @@
+# -*- mode: conf-space; -*-
+# tool/git-riskiness -- estimate of risks of changing files in the MPS tree
+#
+# This file is used by the script tool/git-risks to assess the
+# riskiness of changes made to the MPS.  [Insert why it was developed,
+# ref to issue, etc. RB 2023-02-24]
+#
+# First draft by manual inspection by RB and JPH on 2023-02-20.
+#
+# This file can contain blank lines and comments starting with '#'.
+# See tool/git-risks.
+
+0 0 code/.gitignore
+0 0 code/.p4ignore
+1 0 code/abq.c
+1 0 code/abq.h
+0 0 code/abqtest.c
+0 0 code/airtest.c
+0 0 code/amcss.c
+0 0 code/amcsshe.c
+0 0 code/amcssth.c
+0 0 code/amsss.c
+0 0 code/amssshe.c
+0 0 code/anangc.gmk
+0 0 code/ananll.gmk
+0 0 code/ananmv.nmk
+0 0 code/apss.c
+2 2 code/arena.c
+2 0 code/arenacl.c
+0 0 code/arenacv.c
+2 2 code/arenavm.c
+1 1 code/arg.c
+1 1 code/arg.h
+0 0 code/awlut.c
+0 0 code/awluthe.c
+0 0 code/awlutth.c
+1 1 code/boot.c
+1 1 code/boot.h
+3 3 code/bt.c
+3 3 code/bt.h
+0 0 code/btcv.c
+0 0 code/bttest.c
+4 4 code/buffer.c
+3 3 code/cbs.c
+3 3 code/cbs.h
+? ? chain.h extra file
+3 3 code/check.h
+1 0 code/clock.h
+0 0 code/comm.gmk
+0 0 code/commpost.nmk
+0 0 code/commpre.nmk
+1 1 code/config.h
+2 0 code/dbgpool.c
+2 0 code/dbgpool.h
+2 0 code/dbgpooli.c
+0 0 code/djbench.c
+1 0 code/event.c
+1 0 code/event.h
+0 0 code/eventcnv.c
+1 0 code/eventcom.h
+1 0 code/eventdef.h
+0 0 code/eventpy.c
+0 0 code/eventsql.c
+0 0 code/eventtxt.c
+3 3 code/failover.c
+3 3 code/failover.h
+0 0 code/finalcv.c
+0 0 code/finaltest.c
+0 0 code/fmtdy.c
+0 0 code/fmtdy.h
+0 0 code/fmtdytst.c
+0 0 code/fmtdytst.h
+0 0 code/fmthe.c
+0 0 code/fmthe.h
+0 0 code/fmtno.c
+0 0 code/fmtno.h
+0 0 code/fmtscheme.c
+0 0 code/fmtscheme.h
+0 0 code/forktest.c
+2 2 code/format.c
+0 0 code/fotest.c
+3 3 code/freelist.c
+3 3 code/freelist.h
+0 0 code/fri3gc.gmk
+0 0 code/fri3ll.gmk
+0 0 code/fri6gc.gmk
+0 0 code/fri6ll.gmk
+0 0 code/gc.gmk
+0 0 code/gcbench.c
+0 0 code/gcovfmt.py
+0 0 code/getopt.h
+0 0 code/getoptl.c
+2 2 code/global.c
+0 0 code/gp.gmk
+3 3 code/land.c
+0 0 code/landtest.c
+4 4 code/ld.c
+0 0 code/lia6gc.gmk
+0 0 code/lia6ll.gmk
+0 0 code/lii3gc.gmk
+0 0 code/lii6gc.gmk
+0 0 code/lii6ll.gmk
+0 0 code/ll.gmk
+0 0 code/locbwcss.c
+1 1 code/lock.h
+1 0 code/lockan.c
+0 0 code/lockcov.c
+2 0 code/lockix.c
+0 0 code/lockut.c
+0 0 code/lockutw3.c
+2 2 code/lockw3.c
+3 3 code/locus.c
+3 3 code/locus.h
+0 0 code/locusss.c
+0 0 code/locv.c
+1 1 code/message.c
+0 0 code/messtest.c
+1 1 code/meter.c
+1 1 code/meter.h
+3 3 code/misc.h
+2 2 code/mpm.c
+2 2 code/mpm.h
+0 0 code/mpmss.c
+2 2 code/mpmst.h
+2 2 code/mpmtypes.h
+1 1 code/mps.c
+3 3 code/mps.h
+0 0 code/mps.xcodeproj/project.pbxproj
+2 0 code/mpsacl.h
+2 2 code/mpsavm.h
+1 1 code/mpscamc.h
+1 0 code/mpscams.h
+1 1 code/mpscawl.h
+1 0 code/mpsclo.h
+1 0 code/mpscmfs.h
+1 0 code/mpscmv2.h
+1 0 code/mpscmvff.h
+1 0 code/mpscmvt.h
+1 0 code/mpscsnc.h
+2 2 code/mpsi.c
+0 0 code/mpsicv.c
+1 1 code/mpsio.h
+1 1 code/mpsioan.c
+2 2 code/mpsiw3.c
+1 1 code/mpslib.h
+1 1 code/mpsliban.c
+1 1 code/mpstd.h
+1 1 code/mpswin.h
+0 0 code/mv.nmk
+0 0 code/mv2test.c
+3 3 code/nailboard.c
+3 3 code/nailboard.h
+0 0 code/nailboardtest.c
+0 0 code/pc.nmk
+3 3 code/policy.c
+2 2 code/pool.c
+1 1 code/poolabs.c
+3 3 code/poolamc.c
+3 0 code/poolams.c
+1 0 code/poolams.h
+3 3 code/poolawl.c
+3 0 code/poollo.c
+3 3 code/poolmfs.c
+1 1 code/poolmfs.h
+3 3 code/poolmrg.c
+1 1 code/poolmrg.h
+3 0 code/poolmv2.c
+1 0 code/poolmv2.h
+3 3 code/poolmvff.c
+1 1 code/poolmvff.h
+0 0 code/pooln.c
+0 0 code/pooln.h
+0 0 code/poolncv.c
+3 0 code/poolsnc.c
+3 3 code/prmc.h
+3 0 code/prmcan.c
+3 0 code/prmcanan.c
+3 0 code/prmcfri3.c
+3 0 code/prmcfri6.c
+3 0 code/prmci3.c
+3 0 code/prmci3.h
+3 0 code/prmci3fr.c
+3 0 code/prmci3li.c
+3 0 code/prmci3w3.c
+3 0 code/prmci3xc.c
+3 3 code/prmci6.c
+3 3 code/prmci6.h
+3 0 code/prmci6fr.c
+3 0 code/prmci6li.c
+3 3 code/prmci6w3.c
+3 0 code/prmci6xc.c
+3 0 code/prmcix.c
+3 0 code/prmcix.h
+3 0 code/prmclia6.c
+3 0 code/prmclii3.c
+3 0 code/prmclii6.c
+3 3 code/prmcw3.c
+3 3 code/prmcw3.h
+3 0 code/prmcw3i3.c
+3 3 code/prmcw3i6.c
+3 0 code/prmcxc.c
+3 0 code/prmcxc.h
+3 0 code/prmcxca6.c
+3 0 code/prmcxci3.c
+3 0 code/prmcxci6.c
+1 1 code/prot.h
+1 0 code/protan.c
+2 0 code/proti3.c
+2 2 code/proti6.c
+2 0 code/protix.c
+2 2 code/protocol.c
+2 2 code/protocol.h
+3 0 code/protsgix.c
+3 3 code/protw3.c
+3 0 code/protxc.c
+3 0 code/protxc.h
+2 0 code/pthrdext.c
+2 0 code/pthrdext.h
+0 0 code/qs.c
+2 2 code/range.c
+2 2 code/range.h
+2 2 code/rangetree.c
+2 2 code/rangetree.h
+3 3 code/ref.c
+2 2 code/ring.c
+2 2 code/ring.h
+3 3 code/root.c
+3 3 code/sa.c
+3 3 code/sa.h
+2 0 code/sac.c
+2 0 code/sac.h
+0 0 code/sacss.c
+4 4 code/scan.c
+3 3 code/seg.c
+0 0 code/segsmss.c
+4 4 code/shield.c
+0 0 code/sncss.c
+1 1 code/sp.h
+3 0 code/span.c
+? ? spi3.c
+3 3 code/splay.c
+1 1 code/splay.h
+3 0 code/spw3i3.c
+3 3 code/spw3i6.c
+3 3 code/ss.c
+3 3 code/ss.h
+0 0 code/steptest.c
+2 2 code/table.c
+1 1 code/table.h
+0 0 code/tagtest.c
+0 0 code/teletest.c
+0 0 code/testlib.c
+0 0 code/testlib.h
+0 0 code/testthr.h
+0 0 code/testthrix.c
+0 0 code/testthrw3.c
+1 1 code/th.h
+1 0 code/than.c
+2 0 code/thix.c
+2 2 code/thw3.c
+2 0 code/thxc.c
+4 4 code/trace.c
+3 3 code/traceanc.c
+2 2 code/tract.c
+2 2 code/tract.h
+2 2 code/tree.c
+1 1 code/tree.h
+1 1 code/version.c
+1 1 code/vm.c
+1 1 code/vm.h
+1 1 code/vman.c
+2 2 code/vmix.c
+2 2 code/vmw3.c
+0 0 code/w3i3mv.nmk
+0 0 code/w3i3pc.nmk
+0 0 code/w3i6mv.bat
+0 0 code/w3i6mv.nmk
+0 0 code/w3i6pc.nmk
+0 0 code/w3mv/.gitignore
+0 0 code/w3mv/.p4ignore
+0 0 code/w3mv/amcss/amcss.vcxproj
+0 0 code/w3mv/eventcnv/eventcnv.vcxproj
+0 0 code/w3mv/mps.sln
+0 0 code/w3mv/mps/mps.vcxproj
+3 3 code/walk.c
+0 0 code/walkt0.c
+0 0 code/xca6ll.gmk
+0 0 code/xci3gc.gmk
+0 0 code/xci3ll.gmk
+0 0 code/xci6gc.gmk
+0 0 code/xci6ll.gmk
+0 0 code/zcoll.c
+0 0 code/zmess.c
diff --git a/tool/git-risks b/tool/git-risks
new file mode 100755
index 0000000000..46a003a103
--- /dev/null
+++ b/tool/git-risks
@@ -0,0 +1,37 @@
+#!/bin/bash
+# Estimate the risk of changes to the MPS.
+#
+# [Insert why this was developed, ref to issue, etc.  RB 2023-02-24]
+#
+# Takes arguments to ``git log``.  For example, to assess changes on
+# master::
+#
+#   tool/git-risks --first-parent -n 10 master
+#
+# The estimate of the risks of changing files is stored in
+# tool/git-riskiness
+#
+# TODO: Take option to select risk column argument to sort.
+#
+# Developed from <https://info.ravenbrook.com/mail/2023/02/21/17-08-50/0/>.
+
+git log --format="%H %h %s" "$@" |
+while read -r hash commit subject; do
+    risks=$(
+	# Strip comments and blank lines from git-riskiness
+	sed --regexp-extended \
+	    --expression='s/[[:space:]]*#.*$//; /^[[:space:]]*$/ d' < tool/git-riskiness |
+	# Filter lines that mention files changed in the commit
+	(
+	    grep --fixed-strings \
+		 --file=<(git log --max-count=1 --patch \
+			     --diff-merges=on --name-only \
+			     --pretty="format:" "$hash") ||
+		echo "0 0 no-match"
+	) |
+	# Pick the riskiest
+        sort --numeric --reverse --key=1 |
+	head --lines=1
+    )
+    echo "$commit $risks $subject"
+done

From 68d828280e44fef3f7b910a7408fe26b1e7a72d8 Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 20:31:00 +0000
Subject: [PATCH 2/7] Clarifying what the columns are for.

---
 tool/git-riskiness | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tool/git-riskiness b/tool/git-riskiness
index 8ab6950af2..36174e6924 100644
--- a/tool/git-riskiness
+++ b/tool/git-riskiness
@@ -10,6 +10,8 @@
 # This file can contain blank lines and comments starting with '#'.
 # See tool/git-risks.
 
+# First column is risk to general MPS users
+# Second column is risk to Configura
 0 0 code/.gitignore
 0 0 code/.p4ignore
 1 0 code/abq.c

From 511760fdbdbf8569cf66e3b2e0351622b4c8b3cc Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 20:31:30 +0000
Subject: [PATCH 3/7] For the purpose of risk, ignoring lines that look like
 comments.

---
 tool/git-risks | 35 +++++++++++++++++++++--------------
 1 file changed, 21 insertions(+), 14 deletions(-)

diff --git a/tool/git-risks b/tool/git-risks
index 46a003a103..e06de1251b 100755
--- a/tool/git-risks
+++ b/tool/git-risks
@@ -18,20 +18,27 @@
 git log --format="%H %h %s" "$@" |
 while read -r hash commit subject; do
     risks=$(
-	# Strip comments and blank lines from git-riskiness
+	# Strip comments and blank lines from git-riskiness.
 	sed --regexp-extended \
 	    --expression='s/[[:space:]]*#.*$//; /^[[:space:]]*$/ d' < tool/git-riskiness |
-	# Filter lines that mention files changed in the commit
-	(
-	    grep --fixed-strings \
-		 --file=<(git log --max-count=1 --patch \
-			     --diff-merges=on --name-only \
-			     --pretty="format:" "$hash") ||
-		echo "0 0 no-match"
-	) |
-	# Pick the riskiest
-        sort --numeric --reverse --key=1 |
-	head --lines=1
-    )
-    echo "$commit $risks $subject"
+	    # Filter for lines that mention files changed in the commit.
+	    # Ignores lines that look like comments.
+	    (
+		grep --fixed-strings \
+		     --file=<(git show --pretty="format:" \
+				  --patch \
+				  --diff-merges=first-parent \
+				  --ignore-matching-lines='^[[:space:]]+\*[[:space:]]' \
+				  --ignore-matching-lines='^[[:space:]]*/\*[[:space:]]' \
+				  --ignore-space-change \
+				  "$hash" |
+				  sed --quiet --expression='s!--- a/!!p') ||
+		    echo "0 0 no-match"
+	    ) |
+	    # Pick the riskiest.
+	    # NOTE: --key is 1-based.
+	    sort --numeric --reverse --key=2 |
+	    head --lines=1
+	 )
+    echo "$commit $loc $risks $subject"
 done

From 39ed92fb6ea93d05c348cb182fe062bc8b174313 Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 20:59:52 +0000
Subject: [PATCH 4/7] Output a count of non-comment lines changed.  Remove
 bashism.

---
 tool/git-risks | 35 +++++++++++++++++++++++------------
 1 file changed, 23 insertions(+), 12 deletions(-)

diff --git a/tool/git-risks b/tool/git-risks
index e06de1251b..08ae2eba52 100755
--- a/tool/git-risks
+++ b/tool/git-risks
@@ -15,24 +15,35 @@
 #
 # Developed from <https://info.ravenbrook.com/mail/2023/02/21/17-08-50/0/>.
 
+if difftmp=$(mktemp); then
+    trap "rm -f -- '$difftmp'" EXIT
+else
+    echo 1>&2 "Unable to create temporary file"
+    exit 1
+fi
+    
 git log --format="%H %h %s" "$@" |
 while read -r hash commit subject; do
+    # Get a diff for the change.  Ignoring lines that look like
+    # comments, or with just white space changes.  This is hacky
+    # because it only recognizes C comments.
+    git show --pretty="format:" --patch \
+	--diff-merges=first-parent \
+	--ignore-matching-lines='^[[:space:]]+\*[[:space:]]' \
+	--ignore-matching-lines='^[[:space:]]*/\*[[:space:]]' \
+	--ignore-space-change \
+	"$hash" > $difftmp
+    loc=$(wc -l < "$difftmp")
+    # Edit diff to just a list of files changed
+    sed --in-place --quiet --expression='s!--- a/!!p' "$difftmp"
     risks=$(
 	# Strip comments and blank lines from git-riskiness.
 	sed --regexp-extended \
-	    --expression='s/[[:space:]]*#.*$//; /^[[:space:]]*$/ d' < tool/git-riskiness |
+	    --expression='s/[[:space:]]*#.*$//; /^[[:space:]]*$/ d' \
+	    < tool/git-riskiness |
 	    # Filter for lines that mention files changed in the commit.
-	    # Ignores lines that look like comments.
 	    (
-		grep --fixed-strings \
-		     --file=<(git show --pretty="format:" \
-				  --patch \
-				  --diff-merges=first-parent \
-				  --ignore-matching-lines='^[[:space:]]+\*[[:space:]]' \
-				  --ignore-matching-lines='^[[:space:]]*/\*[[:space:]]' \
-				  --ignore-space-change \
-				  "$hash" |
-				  sed --quiet --expression='s!--- a/!!p') ||
+		grep --fixed-strings --file="$difftmp" ||
 		    echo "0 0 no-match"
 	    ) |
 	    # Pick the riskiest.
@@ -40,5 +51,5 @@ while read -r hash commit subject; do
 	    sort --numeric --reverse --key=2 |
 	    head --lines=1
 	 )
-    echo "$commit $loc $risks $subject"
+    echo "$commit $risks $loc $subject"
 done

From 4b1265a5877db19398dd384800ad890118bbe16e Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 21:09:04 +0000
Subject: [PATCH 5/7] Improving output formatting.  Fixing shellcheck warnings.

---
 tool/git-risks | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tool/git-risks b/tool/git-risks
index 08ae2eba52..86a8770ac0 100755
--- a/tool/git-risks
+++ b/tool/git-risks
@@ -16,6 +16,7 @@
 # Developed from <https://info.ravenbrook.com/mail/2023/02/21/17-08-50/0/>.
 
 if difftmp=$(mktemp); then
+    # shellcheck disable=SC2064
     trap "rm -f -- '$difftmp'" EXIT
 else
     echo 1>&2 "Unable to create temporary file"
@@ -32,7 +33,7 @@ while read -r hash commit subject; do
 	--ignore-matching-lines='^[[:space:]]+\*[[:space:]]' \
 	--ignore-matching-lines='^[[:space:]]*/\*[[:space:]]' \
 	--ignore-space-change \
-	"$hash" > $difftmp
+	"$hash" > "$difftmp"
     loc=$(wc -l < "$difftmp")
     # Edit diff to just a list of files changed
     sed --in-place --quiet --expression='s!--- a/!!p' "$difftmp"
@@ -51,5 +52,5 @@ while read -r hash commit subject; do
 	    sort --numeric --reverse --key=2 |
 	    head --lines=1
 	 )
-    echo "$commit $risks $loc $subject"
+    printf "%s %5d %s %s\n" "$commit" "$loc" "$risks" "$subject"
 done

From 3334f1a1dc379eff1ce1cba6d6945f6e1fd0d293 Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Fri, 24 Feb 2023 21:48:43 +0000
Subject: [PATCH 6/7] Adding option to select risk column.

---
 tool/git-risks | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)

diff --git a/tool/git-risks b/tool/git-risks
index 86a8770ac0..d683632aec 100755
--- a/tool/git-risks
+++ b/tool/git-risks
@@ -11,10 +11,23 @@
 # The estimate of the risks of changing files is stored in
 # tool/git-riskiness
 #
-# TODO: Take option to select risk column argument to sort.
-#
 # Developed from <https://info.ravenbrook.com/mail/2023/02/21/17-08-50/0/>.
 
+usage() {
+    echo 1>&2 "Usage: ${0} [-k column] [-- git log options] [<git log path>...]"
+    exit 1
+}
+
+# Default to general risk (column 1 in git-riskiness)
+key=1
+while getopts k: flag; do
+    case "${flag}" in
+	k) key="${OPTARG}";;
+	*) usage;;
+    esac
+done
+shift $((OPTIND-1))
+
 if difftmp=$(mktemp); then
     # shellcheck disable=SC2064
     trap "rm -f -- '$difftmp'" EXIT
@@ -49,7 +62,7 @@ while read -r hash commit subject; do
 	    ) |
 	    # Pick the riskiest.
 	    # NOTE: --key is 1-based.
-	    sort --numeric --reverse --key=2 |
+	    sort --numeric --reverse --key="$key" |
 	    head --lines=1
 	 )
     printf "%s %5d %s %s\n" "$commit" "$loc" "$risks" "$subject"

From 35b7281c5b4fea14fe3f4841b348afc99fe67bc3 Mon Sep 17 00:00:00 2001
From: Richard Brooksby <rb@ravenbrook.com>
Date: Sat, 25 Feb 2023 07:46:42 +0000
Subject: [PATCH 7/7] Adding draft tool script to show C functions changed
 between commits.

---
 tool/git-fundiff | 50 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100755 tool/git-fundiff

diff --git a/tool/git-fundiff b/tool/git-fundiff
new file mode 100755
index 0000000000..7008980080
--- /dev/null
+++ b/tool/git-fundiff
@@ -0,0 +1,50 @@
+#!/bin/sh
+# tool/git-fundiff -- C functions changed between commits
+#
+# Prints a list of functions that differ in a C source file at the
+# LLVM intermediate code level between two Git commits.
+#
+# NOTE: This will print false positivies because llvm-diff produces
+# them, even when comparing a file with itself, e.g. ``llvm-diff x x``
+
+set -e # exit on error
+
+usage() {
+    echo 1>&2 "Usage: $0 <tree-ish> <tree-ish> <clang source>..."
+    echo 1>&2 "Example: $0 HEAD^ HEAD trace.c"
+    exit 2
+}
+
+if ! which clang llvm-diff-14 > /dev/null; then
+    echo 1>&2 "$0: clang or llvm-diff-14 command not found"
+    exit 2
+fi
+
+if test "$#" -lt 2; then usage; fi
+left="$1"
+right="$2"
+shift 2
+
+tmpdir=$(mktemp --directory)
+# shellcheck disable=SC2064
+trap "rm -rf -- '$tmpdir'" EXIT
+mkdir -p "$tmpdir/left" "$tmpdir/right"
+git archive "$left" code | tar -x -C "$tmpdir/left"
+git archive "$right" code | tar -x -C "$tmpdir/right"
+
+status=0
+while test $# -gt 0; do
+    file="$1"
+    shift
+    (cd "$tmpdir/left/code" &&
+	 clang -S -emit-llvm -o "../../left.ll" "$file")
+    (cd "$tmpdir/right/code" &&
+	 clang -S -emit-llvm -o "../../right.ll" "$file")
+    # Try to reduce false positives where the output is identical
+    cmp "$tmpdir/left.ll" "$tmpdir/right.ll" && continue
+    llvm-diff-14 "$tmpdir/left.ll" "$tmpdir/right.ll" \
+		 > "$tmpdir/diff" 2>&1 && continue
+    status=1
+    sed -ne 's/^in function \(.*\):$/\1/p' < "$tmpdir/diff"
+done
+exit "$status"