Skip to contentSkip to conversation

local VAR="VAL"

PATCH In review Junio C Hamano 6 participants 23 messages Apr 6, 2024 – Apr 8, 2024 Documentation

lore
Share & tools
lore
#0 lore raw

cover local VAR="VAL"

* Update coding guidelines and test-lint script so that we quote the right hand side of assignment used with "local", which is buggy on certain versions of dash.

The first patch is not about the theme of the topic, but to document a rule enforced by the test-lint script that is not written down in the coding guidelines.

The second patch gives guidance to avoid the dash bug.

Patches [3/6], [4/6], and [5/6] are to adjust the existing tests. I think many of them are currently safe because the values they assign are $IFS safe, but some may be real workarounds for the dash bug.

The last patch introduces the test-lint pattern.

Junio C Hamano (6):
CodingGuidelines: describe "export VAR=VAL" rule
CodingGuidelines: quote assigned value in 'local var=$val'
t: local VAR="VAL" (quote positional parameters)
t: local VAR="VAL" (quote command substitution)
t: local VAR="VAL" (quote ${magic-reference})
t: teach lint that RHS of 'local VAR=VAL' needs to be quoted

 Documentation/CodingGuidelines | 20 ++++++++++++++++++++
 t/check-non-portable-shell.pl  |  2 ++
 t/lib-parallel-checkout.sh     |  2 +-
 t/t2400-worktree-add.sh        |  2 +-
 t/t4011-diff-symlink.sh        |  4 ++--
 t/t4210-log-i18n.sh            |  4 ++--
 t/test-lib-functions.sh        | 12 ++++++------
 7 files changed, 34 insertions(+), 12 deletions(-)
Signature · 1 line
2.44.0-501-g19981daefd
#1 lore raw

PATCH 1/6 CodingGuidelines: describe "export VAR=VAL" rule

https://lore.kernel.org/git/[email protected]/
resulted in 9968ffff (test-lint: detect 'export FOO=bar', 2013-07-08) to add a rule to t/check-non-portable-shell.pl script to reject

export VAR=VAL

and suggest us to instead write it as "export VAR" followed by "VAR=VAL". This however was not spelled out in the CodingGuidelines document.

We may want to re-evaluate the rule since it is from ages ago, but for now, let's make the written rule and what the automation enforces consistent.

---
 Documentation/CodingGuidelines | 6 ++++++
 1 file changed, 6 insertions(+)
Documentation/CodingGuidelines +6 -0
index 9495df835d..0a39205c48 100644
@@ -188,6 +188,12 @@ For shell scripts specifically (not exhaustive):
    hopefully nobody starts using "local" before they are reimplemented
    in C ;-)
 
+ - Some versions of shell do not understand "export variable=value",
+   so we write "export variable" and "variable=value" on separae
+   lines.  Note that this was reported in 2013 and the situation might
+   have changed since then.  We'd need to re-evaluate this rule,
+   together with the rule in t/check-non-portable-shell.pl script.
+
  - Use octal escape sequences (e.g. "\302\242"), not hexadecimal (e.g.
    "\xc2\xa2") in printf format strings, since hexadecimal escape
    sequences are not portable.
Signature · 1 line
2.44.0-501-g19981daefd
#2 lore raw

Re: [PATCH 1/6] CodingGuidelines: describe "export VAR=VAL" rule

Quoted 10 lines · Junio C Hamano

https://lore.kernel.org/git/[email protected]/
resulted in 9968ffff (test-lint: detect 'export FOO=bar', 2013-07-08) to add a rule to t/check-non-portable-shell.pl script to reject

export VAR=VAL

and suggest us to instead write it as "export VAR" followed by "VAR=VAL". This however was not spelled out in the CodingGuidelines document.

I suspect you meant:

... and suggest us to instead write it as "VAR=VAL" followed by "export VAR".

Documentation/CodingGuidelines For shell scripts specifically (not exhaustive): · line 192

Quoted 13 lines

We may want to re-evaluate the rule since it is from ages ago, but for now, let's make the written rule and what the automation enforces consistent.

---

Documentation/CodingGuidelines +2 -0
@@ -188,6 +188,12 @@ For shell scripts specifically (not exhaustive):
    hopefully nobody starts using "local" before they are reimplemented
    in C ;-)
 
+ - Some versions of shell do not understand "export variable=value",
+   so we write "export variable" and "variable=value" on separae

s/separae/separate/

Here too, it might be clearer to swap around the pieces:

... so we write "variable=value" and "export variable" on...

Documentation/CodingGuidelines For shell scripts specifically (not exhaustive): · line 195

Quoted 3 lines
Documentation/CodingGuidelines +3 -0
+   lines.  Note that this was reported in 2013 and the situation might
+   have changed since then.  We'd need to re-evaluate this rule,
+   together with the rule in t/check-non-portable-shell.pl script.

The bit starting at "Note that..." seems more appropriate for the commit message (which is already the case) or a To-Do list. People reading this document are likely newcomers looking for concrete instructions about how to code for this project, and this sort of To-Do item isn't going to help them. (If anything, it might confuse them into ignoring the advice to split export foo=bar into two statements, which will result in reviewers asking them to reroll.)

#3 lore raw

Re: [PATCH 1/6] CodingGuidelines: describe "export VAR=VAL" rule

Quoted 8 lines · Eric Sunshine

Documentation/CodingGuidelines For shell scripts specifically (not exhaustive): · line 195

Quoted 3 lines
Documentation/CodingGuidelines +3 -0
+   lines.  Note that this was reported in 2013 and the situation might
+   have changed since then.  We'd need to re-evaluate this rule,
+   together with the rule in t/check-non-portable-shell.pl script.

The bit starting at "Note that..." seems more appropriate for the commit message (which is already the case) or a To-Do list. People reading this document are likely newcomers looking for concrete instructions about how to code for this project,...

Very true. I thought I'd move some to the log message, but it turns out that enough is already described there.

Thanks.

#4 lore raw

Re: [PATCH 1/6] CodingGuidelines: describe "export VAR=VAL" rule

In reply to Eric Sunshine
Quoted 8 lines · Eric Sunshine
Quoted 7 lines · Junio C Hamano

https://lore.kernel.org/git/[email protected]/
resulted in 9968ffff (test-lint: detect 'export FOO=bar', 2013-07-08) to add a rule to t/check-non-portable-shell.pl script to reject

export VAR=VAL

Quoted 8 lines
Quoted 3 lines

and suggest us to instead write it as "export VAR" followed by "VAR=VAL". This however was not spelled out in the CodingGuidelines document.

I suspect you meant:

... and suggest us to instead write it as "VAR=VAL" followed by "export VAR".

There is no difference between them. The export command only marks the variable for export, independent of the current or future value of the variable. The exported value is always the last assigned one.

Signature · 3 lines
Andreas Schwab, [email protected]
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."
#5 lore raw

Re: [PATCH 1/6] CodingGuidelines: describe "export VAR=VAL" rule

Quoted 8 lines · Andreas Schwab
Quoted 4 lines

I suspect you meant:

... and suggest us to instead write it as "VAR=VAL" followed by "export VAR".

There is no difference between them. The export command only marks the variable for export, independent of the current or future value of the variable. The exported value is always the last assigned one.

Correct.

But we are talking about working around sub-standard (read: buggy) implementations and it is of dubious value to assume a compliant implementation when devising a workaround.

It is easily imaginable that a sub-standard implementation uses a symbol table with a single "is it exported?" bit in addition to (name, value), without a way to say "this parameter is not set (yet)" (IOW, never value==NULL), and such an implementation would not be capable to have "this name is exported but nobody set the value to it yet". Using an assignment to make sure it is known before setting the exported bit is safer to protect against such an implementation.

#6 lore raw

Re: [PATCH 1/6] CodingGuidelines: describe "export VAR=VAL" rule

In reply to Andreas Schwab
Quoted 13 lines · Andreas Schwab
Quoted 12 lines · Eric Sunshine
Quoted 10 lines · Junio C Hamano

https://lore.kernel.org/git/[email protected]/
resulted in 9968ffff (test-lint: detect 'export FOO=bar', 2013-07-08) to add a rule to t/check-non-portable-shell.pl script to reject

export VAR=VAL

and suggest us to instead write it as "export VAR" followed by "VAR=VAL". This however was not spelled out in the CodingGuidelines document.

Quoted 8 lines
Quoted 4 lines

I suspect you meant:

... and suggest us to instead write it as "VAR=VAL" followed by "export VAR".

There is no difference between them. The export command only marks the variable for export, independent of the current or future value of the variable. The exported value is always the last assigned one.

Yes, I know, but it is customary in this code-base to write it as:

    VAR=VAL &&
    export VAR

not the other way around, so it makes sense for CodingGuidelines to illustrate it in a fashion consistent with its use in the project.

#7 lore raw

PATCH 2/6 CodingGuidelines: quote assigned value in 'local var=$val'

In reply to Junio C Hamano

Dash bug https://bugs.launchpad.net/ubuntu/+source/dash/+bug/139097
lets the shell erroneously perform field splitting on the expansion of a command substitution during declaration of a local or an extern variable.

The explanation was stolen from ebee5580 (parallel-checkout: avoid dash local bug in tests, 2021-06-06).

---
 Documentation/CodingGuidelines | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
Documentation/CodingGuidelines +14 -0
index 0a39205c48..1cb77a871b 100644
@@ -194,6 +194,20 @@ For shell scripts specifically (not exhaustive):
    have changed since then.  We'd need to re-evaluate this rule,
    together with the rule in t/check-non-portable-shell.pl script.
 
+ - Some versions of dash have broken variable assignment when prefixed
+   with "local", "export", and "readonly", in that the value to be
+   assigned goes through field splitting at $IFS unless quoted.
+
+   DO NOT write:
+
+     local variable=$value           ;# wrong
+     local variable=$(command args)  ;# wrong
+
+   and instead write:
+
+     local variable="$value"
+     local variable="$(command args)"
+
  - Use octal escape sequences (e.g. "\302\242"), not hexadecimal (e.g.
    "\xc2\xa2") in printf format strings, since hexadecimal escape
    sequences are not portable.
Signature · 1 line
2.44.0-501-g19981daefd
#8 lore raw

RE: [PATCH 2/6] CodingGuidelines: quote assigned value in 'local var=$val'

Quoted 2 lines · Junio C Hamano

Dash bug https://bugs.launchpad.net/ubuntu/+source/dash/+bug/139097
lets the shell erroneously perform field splitting on the expansion of a

command

Quoted 3 lines

substitution during declaration of a local or an extern variable.

The explanation was stolen from ebee5580 (parallel-checkout: avoid dash

local bug

Review

Quoted 8 lines

in tests, 2021-06-06).

---
Documentation/CodingGuidelines | 14 ++++++++++++++

1 file changed, 14 insertions(+)
(quoted diff) +0 -0

b/Documentation/CodingGuidelines

Quoted 14 lines

index 0a39205c48..1cb77a871b 100644

Documentation/CodingGuidelines +7 -0
@@ -194,6 +194,20 @@ For shell scripts specifically (not exhaustive):
   have changed since then.  We'd need to re-evaluate this rule,
   together with the rule in t/check-non-portable-shell.pl script.
 
+ - Some versions of dash have broken variable assignment when prefixed
+   with "local", "export", and "readonly", in that the value to be
+   assigned goes through field splitting at $IFS unless quoted.
+
+   DO NOT write:
+
+     local variable=$value           ;# wrong

Documentation/CodingGuidelines For shell scripts specifically (not exhaustive): · line 213

Quoted 10 lines
Documentation/CodingGuidelines +7 -0
+     local variable=$(command args)  ;# wrong
+
+   and instead write:
+
+     local variable="$value"
+     local variable="$(command args)"
+
 - Use octal escape sequences (e.g. "\302\242"), not hexadecimal (e.g.
   "\xc2\xa2") in printf format strings, since hexadecimal escape
   sequences are not portable.

I can confirm, at least for the set of platforms I work on, that printf with hex values is definitely not portable.
--Randall

#9 lore raw

Re: [PATCH 2/6] CodingGuidelines: quote assigned value in 'local var=$val'

Quoted 6 lines · <[email protected]> writes:
Quoted 3 lines

- Use octal escape sequences (e.g. "\302\242"), not hexadecimal (e.g. "\xc2\xa2") in printf format strings, since hexadecimal escape sequences are not portable.

I can confirm, at least for the set of platforms I work on, that printf with hex values is definitely not portable.

Sure, that is why we have that rule that we see in the context.

#10 lore raw

Re: [PATCH 2/6] CodingGuidelines: quote assigned value in 'local var=$val'

In reply to Junio C Hamano
Quoted 16 lines · Junio C Hamano

Dash bug https://bugs.launchpad.net/ubuntu/+source/dash/+bug/139097
lets the shell erroneously perform field splitting on the expansion of a command substitution during declaration of a local or an extern variable.

The explanation was stolen from ebee5580 (parallel-checkout: avoid dash local bug in tests, 2021-06-06).

---

Documentation/CodingGuidelines +4 -0
@@ -194,6 +194,20 @@ For shell scripts specifically (not exhaustive):
+ - Some versions of dash have broken variable assignment when prefixed
+   with "local", "export", and "readonly", in that the value to be
+   assigned goes through field splitting at $IFS unless quoted.
+

Documentation/CodingGuidelines For shell scripts specifically (not exhaustive): · line 207

Quoted 10 lines
Documentation/CodingGuidelines +10 -0
+   DO NOT write:
+
+     local variable=$value           ;# wrong
+     local variable=$(command args)  ;# wrong
+
+   and instead write:
+
+     local variable="$value"
+     local variable="$(command args)"
+

Every other example in the shell-script section of this document is written like this:

    (incorrect)
    local variable=$value
    local variable=$(command args)

    (correct)
    local variable="$value"
    local variable="$(command args)"

Should this patch follow suit for consistency?

#11 lore raw

Re: [PATCH 2/6] CodingGuidelines: quote assigned value in 'local var=$val'

Quoted 11 lines · Eric Sunshine

Every other example in the shell-script section of this document is written like this:

    (incorrect)
    local variable=$value
    local variable=$(command args)

    (correct)
    local variable="$value"
    local variable="$(command args)"
Quoted 1 line

Should this patch follow suit for consistency?

Sure. That sounds like a good idea.

#12 lore raw

PATCH 3/6 t: local VAR="VAL" (quote positional parameters)

In reply to Junio C Hamano

Future-proof test scripts that do

local VAR=VAL

without quoting VAL (which is OK in POSIX but broken in some shells) that is a positional parameter, e.g. $4.

---
 t/lib-parallel-checkout.sh | 2 +-
 t/t2400-worktree-add.sh    | 2 +-
 t/t4210-log-i18n.sh        | 4 ++--
 t/test-lib-functions.sh    | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)
t/lib-parallel-checkout.sh +1 -1
index acaee9cbb6..8324d6c96d 100644
@@ -20,7 +20,7 @@ test_checkout_workers () {
 		BUG "too few arguments to test_checkout_workers"
 	fi &&
 
-	local expected_workers=$1 &&
+	local expected_workers="$1" &&
 	shift &&
 
 	local trace_file=trace-test-checkout-workers &&
t/t2400-worktree-add.sh +1 -1
index 051363acbb..5851e07290 100755
@@ -404,7 +404,7 @@ test_expect_success '"add" worktree with orphan branch, lock, and reason' '
 # Note: Quoted arguments containing spaces are not supported.
 test_wt_add_orphan_hint () {
 	local context="$1" &&
-	local use_branch=$2 &&
+	local use_branch="$2" &&
 	shift 2 &&
 	local opts="$*" &&
 	test_expect_success "'worktree add' show orphan hint in bad/orphan HEAD w/ $context" '
t/t4210-log-i18n.sh +2 -2
index d2dfcf164e..75216f19ce 100755
@@ -64,7 +64,7 @@ test_expect_success 'log --grep does not find non-reencoded values (latin1)' '
 '
 
 triggers_undefined_behaviour () {
-	local engine=$1
+	local engine="$1"
 
 	case $engine in
 	fixed)
@@ -85,7 +85,7 @@ triggers_undefined_behaviour () {
 }
 
 mismatched_git_log () {
-	local pattern=$1
+	local pattern="$1"
 
 	LC_ALL=$is_IS_locale git log --encoding=ISO-8859-1 --format=%s \
 		--grep=$pattern
t/test-lib-functions.sh +1 -1
index 2f8868caa1..3204afbafb 100644
@@ -1689,7 +1689,7 @@ test_parse_ls_tree_oids () {
 # Choose a port number based on the test script's number and store it in
 # the given variable name, unless that variable already contains a number.
 test_set_port () {
-	local var=$1 port
+	local var="$1" port
 
 	if test $# -ne 1 || test -z "$var"
 	then
Signature · 1 line
2.44.0-501-g19981daefd
#13 lore raw

Re: [PATCH 3/6] t: local VAR="VAL" (quote positional parameters)

Quoted 20 lines · Junio C Hamano

Future-proof test scripts that do

local VAR=VAL

without quoting VAL (which is OK in POSIX but broken in some shells) that is a positional parameter, e.g. $4.

---
 t/lib-parallel-checkout.sh | 2 +-
 t/t2400-worktree-add.sh    | 2 +-
 t/t4210-log-i18n.sh        | 4 ++--
 t/test-lib-functions.sh    | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)
t/lib-parallel-checkout.sh +0 -0
index acaee9cbb6..8324d6c96d 100644

t/lib-parallel-checkout.sh test_checkout_workers () { · line 24

Quoted 6 lines
t/lib-parallel-checkout.sh +1 -1
@@ -20,7 +20,7 @@ test_checkout_workers () {
 		BUG "too few arguments to test_checkout_workers"
 	fi &&
 
-	local expected_workers=$1 &&
+	local expected_workers="$1" &&
 	shift &&

I was wondering a bit why this is a problem in t0610, but not over here. As far as I understand it these statements are fine in practice because the expanded values cannot be split, right? So if "$1" expanded to something with spaces in between things would start to break.

In any case, changing all of these to be quoted feels like the right thing to do regardless of whether or not it happens to work with the current values of "$1". Otherwise it's simply a confusing failure waiting to happen.

Patrick

Quoted 53 lines
t/lib-parallel-checkout.sh +0 -0
 	local trace_file=trace-test-checkout-workers &&
t/t2400-worktree-add.sh +1 -1
index 051363acbb..5851e07290 100755
@@ -404,7 +404,7 @@ test_expect_success '"add" worktree with orphan branch, lock, and reason' '
 # Note: Quoted arguments containing spaces are not supported.
 test_wt_add_orphan_hint () {
 	local context="$1" &&
-	local use_branch=$2 &&
+	local use_branch="$2" &&
 	shift 2 &&
 	local opts="$*" &&
 	test_expect_success "'worktree add' show orphan hint in bad/orphan HEAD w/ $context" '
t/t4210-log-i18n.sh +2 -2
index d2dfcf164e..75216f19ce 100755
@@ -64,7 +64,7 @@ test_expect_success 'log --grep does not find non-reencoded values (latin1)' '
 '
 
 triggers_undefined_behaviour () {
-	local engine=$1
+	local engine="$1"
 
 	case $engine in
 	fixed)
@@ -85,7 +85,7 @@ triggers_undefined_behaviour () {
 }
 
 mismatched_git_log () {
-	local pattern=$1
+	local pattern="$1"
 
 	LC_ALL=$is_IS_locale git log --encoding=ISO-8859-1 --format=%s \
 		--grep=$pattern
t/test-lib-functions.sh +1 -1
index 2f8868caa1..3204afbafb 100644
@@ -1689,7 +1689,7 @@ test_parse_ls_tree_oids () {
 # Choose a port number based on the test script's number and store it in
 # the given variable name, unless that variable already contains a number.
 test_set_port () {
-	local var=$1 port
+	local var="$1" port
 
 	if test $# -ne 1 || test -z "$var"
 	then
Signature · 1 line
2.44.0-501-g19981daefd
#14 lore raw

Re: [PATCH 3/6] t: local VAR="VAL" (quote positional parameters)

Quoted 24 lines · Patrick Steinhardt
Quoted 23 lines · Junio C Hamano

Future-proof test scripts that do

local VAR=VAL

without quoting VAL (which is OK in POSIX but broken in some shells) that is a positional parameter, e.g. $4.

---
 t/lib-parallel-checkout.sh | 2 +-
 t/t2400-worktree-add.sh    | 2 +-
 t/t4210-log-i18n.sh        | 4 ++--
 t/test-lib-functions.sh    | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)
t/lib-parallel-checkout.sh +0 -0
index acaee9cbb6..8324d6c96d 100644
@@ -20,7 +20,7 @@ test_checkout_workers () {
 		BUG "too few arguments to test_checkout_workers"
 	fi &&
 
Quoted 8 lines

t/lib-parallel-checkout.sh test_checkout_workers () { · line 24

Quoted 3 lines
t/lib-parallel-checkout.sh +1 -1
-	local expected_workers=$1 &&
+	local expected_workers="$1" &&
 	shift &&

I was wondering a bit why this is a problem in t0610, but not over here. As far as I understand it these statements are fine in practice because the expanded values cannot be split, right? So if "$1" expanded to something with spaces in between things would start to break.

Correct.

Quoted 4 lines

In any case, changing all of these to be quoted feels like the right thing to do regardless of whether or not it happens to work with the current values of "$1". Otherwise it's simply a confusing failure waiting to happen.

Again, agreed. That is where my "Future-proof" comes from.

The true objective of this change is so that the last patch does not have to learn too much exceptions ;-) As long as expected_workers is expected to be a number (an unbroken sequence of digits), even if we add more callers to this helper in the future, $1 we see here is expected to be $IFS safe. So in that sense my "future-proof" is a white lie.

#15 lore raw

PATCH 4/6 t: local VAR="VAL" (quote command substitution)

In reply to Junio C Hamano

Future-proof test scripts that do

local VAR=VAL

without quoting VAL (which is OK in POSIX but broken in some shells) that is a $(command substitution).

---
 t/t4011-diff-symlink.sh | 4 ++--
 t/test-lib-functions.sh | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)
t/t4011-diff-symlink.sh +2 -2
index d7a5f7ae78..bc8ba88719 100755
@@ -13,13 +13,13 @@ TEST_PASSES_SANITIZE_LEAK=true
 
 # Print the short OID of a symlink with the given name.
 symlink_oid () {
-	local oid=$(printf "%s" "$1" | git hash-object --stdin) &&
+	local oid="$(printf "%s" "$1" | git hash-object --stdin)" &&
 	git rev-parse --short "$oid"
 }
 
 # Print the short OID of the given file.
 short_oid () {
-	local oid=$(git hash-object "$1") &&
+	local oid="$(git hash-object "$1")" &&
 	git rev-parse --short "$oid"
 }
 
t/test-lib-functions.sh +1 -1
index 3204afbafb..3dc638f7dc 100644
@@ -1764,7 +1764,7 @@ test_subcommand () {
 		shift
 	fi
 
-	local expr=$(printf '"%s",' "$@")
+	local expr="$(printf '"%s",' "$@")"
 	expr="${expr%,}"
 
 	if test -n "$negate"
Signature · 1 line
2.44.0-501-g19981daefd
#16 lore raw

PATCH 5/6 t: local VAR="VAL" (quote ${magic-reference})

In reply to Junio C Hamano

Future-proof test scripts that do

local VAR=VAL

without quoting VAL (which is OK in POSIX but broken in some shells) that is ${magic-"reference to a parameter"}.

---
 t/test-lib-functions.sh | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
t/test-lib-functions.sh +4 -4
index 3dc638f7dc..029cb31ffe 100644
@@ -330,7 +330,7 @@ test_commit () {
 		shift
 	done &&
 	indir=${indir:+"$indir"/} &&
-	local file=${2:-"$1.t"} &&
+	local file="${2:-"$1.t"}" &&
 	if test -n "$append"
 	then
 		$echo "${3-$1}" >>"$indir$file"
@@ -1672,7 +1672,7 @@ test_oid () {
 # Insert a slash into an object ID so it can be used to reference a location
 # under ".git/objects".  For example, "deadbeef..." becomes "de/adbeef..".
 test_oid_to_path () {
-	local basename=${1#??}
+	local basename="${1#??}"
 	echo "${1%$basename}/$basename"
 }
 
@@ -1840,7 +1840,7 @@ test_readlink () {
 # An optional increment to the magic timestamp may be specified as second
 # argument.
 test_set_magic_mtime () {
-	local inc=${2:-0} &&
+	local inc="${2:-0}" &&
 	local mtime=$((1234567890 + $inc)) &&
 	test-tool chmtime =$mtime "$1" &&
 	test_is_magic_mtime "$1" $inc
@@ -1853,7 +1853,7 @@ test_set_magic_mtime () {
 # argument.  Usually, this should be the same increment which was used for
 # the associated test_set_magic_mtime.
 test_is_magic_mtime () {
-	local inc=${2:-0} &&
+	local inc="${2:-0}" &&
 	local mtime=$((1234567890 + $inc)) &&
 	echo $mtime >.git/test-mtime-expect &&
 	test-tool chmtime --get "$1" >.git/test-mtime-actual &&
Signature · 1 line
2.44.0-501-g19981daefd
#17 lore raw

PATCH 6/6 t: teach lint that RHS of 'local VAR=VAL' needs to be quoted

In reply to Junio C Hamano

Teach t/check-non-portable-shell.pl that right hand side of the assignment done with "local VAR=VAL" need to be quoted. We deliberately target only VAL that begins with $ so that we can catch

- $variable_reference and positional parameter reference like $4
- $(command substitution)
- ${variable_reference-with_magic}

while excluding

- $'\n' that is a bash-ism freely usable in t990[23]
- $(( arithmetic )) whose result should be $IFS safe.
- $? that also is $IFS safe

---
 t/check-non-portable-shell.pl | 2 ++
 1 file changed, 2 insertions(+)
t/check-non-portable-shell.pl +2 -0
index dd8107cd7d..b2b28c2ced 100755
@@ -47,6 +47,8 @@ sub err {
 	/\bgrep\b.*--file\b/ and err 'grep --file FILE is not portable (use grep -f FILE)';
 	/\b[ef]grep\b/ and err 'egrep/fgrep obsolescent (use grep -E/-F)';
 	/\bexport\s+[A-Za-z0-9_]*=/ and err '"export FOO=bar" is not portable (use FOO=bar && export FOO)';
+	/\blocal\s+[A-Za-z0-9_]*=\$([A-Za-z0-9_{]|[(][^(])/ and
+		err q(quote "$val" in 'local var=$val');
 	/^\s*([A-Z0-9_]+=(\w*|(["']).*?\3)\s+)+(\w+)/ and exists($func{$4}) and
 		err '"FOO=bar shell_func" assignment extends beyond "shell_func"';
 	$line = '';
Signature · 1 line
2.44.0-501-g19981daefd
#18 lore raw

Re: [PATCH 6/6] t: teach lint that RHS of 'local VAR=VAL' needs to be quoted

Quoted 8 lines · Junio C Hamano

Teach t/check-non-portable-shell.pl that right hand side of the assignment done with "local VAR=VAL" need to be quoted. We deliberately target only VAL that begins with $ so that we can catch

- $variable_reference and positional parameter reference like $4
- $(command substitution)
- ${variable_reference-with_magic}

Quoted 5 lines

while excluding

- $'\n' that is a bash-ism freely usable in t990[23]
- $(( arithmetic )) whose result should be $IFS safe.
- $? that also is $IFS safe

Hmm. Just porting over my comment from the other thread (before I realized you'd written this series), this misses:

local foo=bar/$1

etc. Should we look for the "$" anywhere on the line? I doubt we can get things foolproof, but requiring somebody to quote:

  local foo=$((1+2))

does not seem like the worst outcome. I dunno.

-Peff

#19 lore raw

Re: [PATCH 6/6] t: teach lint that RHS of 'local VAR=VAL' needs to be quoted

Quoted 19 lines · Jeff King
Quoted 8 lines · Junio C Hamano

Teach t/check-non-portable-shell.pl that right hand side of the assignment done with "local VAR=VAL" need to be quoted. We deliberately target only VAL that begins with $ so that we can catch

- $variable_reference and positional parameter reference like $4
- $(command substitution)
- ${variable_reference-with_magic}

Quoted 5 lines

while excluding

- $'\n' that is a bash-ism freely usable in t990[23]
- $(( arithmetic )) whose result should be $IFS safe.
- $? that also is $IFS safe

Hmm. Just porting over my comment from the other thread (before I realized you'd written this series), this misses:

Quoted 8 lines

local foo=bar/$1

etc. Should we look for the "$" anywhere on the line? I doubt we can get things foolproof, but requiring somebody to quote:

  local foo=$((1+2))

does not seem like the worst outcome. I dunno.

Looking at the output from

    $ git grep -E -e 'local [a-zA-Z0-9_]+=[^"]*[$]' t/

the listed ones in the proposed commit log message are the false positives. Luckily we didn't have anything that tries to concatenate parameter reference to something else.

But with the pattern we do miss

local var=$*

and possibly many others. So I am not sure. The false positives do look moderately bad, so I'd rather start with the simplest one proposed in the patch.

#20 lore raw

Re: [PATCH 6/6] t: teach lint that RHS of 'local VAR=VAL' needs to be quoted

Quoted 20 lines · Junio C Hamano
Quoted 11 lines

Hmm. Just porting over my comment from the other thread (before I realized you'd written this series), this misses:

local foo=bar/$1

etc. Should we look for the "$" anywhere on the line? I doubt we can get things foolproof, but requiring somebody to quote:

  local foo=$((1+2))

does not seem like the worst outcome. I dunno.

Looking at the output from

    $ git grep -E -e 'local [a-zA-Z0-9_]+=[^"]*[$]' t/

the listed ones in the proposed commit log message are the false positives. Luckily we didn't have anything that tries to concatenate parameter reference to something else.

Quoted 7 lines

But with the pattern we do miss

local var=$*

and possibly many others. So I am not sure. The false positives do look moderately bad, so I'd rather start with the simplest one proposed in the patch.

Yeah, I think a regex is probably going to end up with either false positives or false negatives. It probably does not matter too much which way we err, if we expect them to be rare on either side.

My thinking was mostly that false negatives are worse, because they only matter on old buggy versions of dash (and only if the tests actually pass a value with spaces). And so most developers will not notice them immediately. Whereas false positives, while annoying, are reported to them immediately by the linter. And generally, dealing with problems closer to the time of writing means less work overall.

But I am happy to take your series as-is and we can see which cases (if any!) we miss in practice.

I do hope that eventually we could just say "that buggy version of dash does not matter anymore", but I think it is too soon for that (it sounds like it is still being used in CI).

-Peff

#21 lore raw

[PATCH 7/6] t0610: local VAR="VAL" fix

In reply to Junio C Hamano

The series was based on maint and fixes all the tests that exist there, but we have acquired a few more.

---
 t/t0610-reftable-basics.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
t/t0610-reftable-basics.sh +1 -1
index 686781192e..c8074ebab2 100755
@@ -83,7 +83,7 @@ test_expect_success 'init: reinitializing reftable with files backend fails' '
 test_expect_perms () {
 	local perms="$1"
 	local file="$2"
-	local actual=$(ls -l "$file") &&
+	local actual="$(ls -l "$file")" &&
 
 	case "$actual" in
 	$perms*)
#22 lore raw

[PATCH 8/6] t1016: local VAR="VAL" fix

In reply to Junio C Hamano

The series was based on maint and fixes all the tests that exist there, but we have acquired a few more.

I suspect that the values assigned in many of these places are $IFS safe, and this is primarily to squelch the linter than adding a necessary workaround for buggy dash.

---
 t/t1016-compatObjectFormat.sh | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
t/t1016-compatObjectFormat.sh +6 -6
index 8132cd37b8..be3206a16f 100755
@@ -79,7 +79,7 @@ commit2_oid () {
 }
 
 del_sigcommit () {
-    local delete=$1
+    local delete="$1"
 
     if test "$delete" = "sha256" ; then
 	local pattern="gpgsig-sha256"
@@ -91,8 +91,8 @@ del_sigcommit () {
 
 
 del_sigtag () {
-    local storage=$1
-    local delete=$2
+    local storage="$1"
+    local delete="$2"
 
     if test "$storage" = "$delete" ; then
 	local pattern="trailer"
@@ -181,7 +181,7 @@ done
 cd "$base"
 
 compare_oids () {
-    test "$#" = 5 && { local PREREQ=$1; shift; } || PREREQ=
+    test "$#" = 5 && { local PREREQ="$1"; shift; } || PREREQ=
     local type="$1"
     local name="$2"
     local sha1_oid="$3"
@@ -193,8 +193,8 @@ compare_oids () {
 
     git --git-dir=repo-sha1/.git rev-parse --output-object-format=sha256 ${sha1_oid} > ${name}_sha1_sha256_found
     git --git-dir=repo-sha256/.git rev-parse --output-object-format=sha1 ${sha256_oid} > ${name}_sha256_sha1_found
-    local sha1_sha256_oid=$(cat ${name}_sha1_sha256_found)
-    local sha256_sha1_oid=$(cat ${name}_sha256_sha1_found)
+    local sha1_sha256_oid="$(cat ${name}_sha1_sha256_found)"
+    local sha256_sha1_oid="$(cat ${name}_sha256_sha1_found)"
 
     test_expect_success $PREREQ "Verify ${type} ${name}'s sha1 oid" '
 	git --git-dir=repo-sha256/.git rev-parse --output-object-format=sha1 ${sha256_oid} > ${name}_sha1 &&