Build tags assigned incorrectly ("netgo ledger,") with Make 4.3+

[sigv]

Summary of Bug

Appending More Text to Variables is done in Makefile when working with build tags.
Something changed in Make 4.3 regarding empty values being appended to themselves using the += operator.

Ubuntu 20.04 LTS ships with Make 4.2 and results in build tags "netgo,ledger".
Ubuntu 22.04 LTS ships with Make 4.3 and results in build tags "netgo ledger,".

Quoting the Make documentation:

Using ‘+=’ is similar to:

objects = main.o foo.o bar.o utils.o
objects := $(objects) another.o

but differs in ways that become important when you use more complex values.

Version

Any project version, when building with Make 4.3 or Make 4.4.

Steps to Reproduce

Tested on Ubuntu 22.04.1, however should apply to any distro/version as we are building upstream Make for test:

sudo apt -qy install make gcc

WORKSPACE=$(mktemp -d --tmpdir make.XXX)
cd "$WORKSPACE"

# Create minimal Makefile based on logic in repository

cat >Makefile <<EOF
build_tags = netgo
build_tags += ledger

whitespace :=
whitespace += \$(whitespace)
comma := ,
build_tags_comma_sep := \$(subst \$(whitespace),\$(comma),\$(build_tags))

.PHONY: all
all:
$(echo -e '\t')@echo '"\$(build_tags_comma_sep)"'
EOF

# Additional Makefile that uses explicit assignment

cp Makefile Makefile.proposal
sed -i 's/^whitespace += \$(whitespace)$/whitespace := $(whitespace) $(whitespace)/' Makefile.proposal

# Build relevant Make versions for testing

wget -nv https://ftp.gnu.org/gnu/make/make-4.2.tar.gz && tar -xf make-4.2.tar.gz
# Make 4.2 does not play nice with newer releases. Needs a patch to check "newer or equal" instead of just "equal".
sed -i 's/^# if _GNU_GLOB_INTERFACE_VERSION == GLOB_INTERFACE_VERSION$/# if _GNU_GLOB_INTERFACE_VERSION >= GLOB_INTERFACE_VERSION/' make-4.2/glob/glob.c
(cd make-4.2 && ./configure && make)

wget -nv https://ftp.gnu.org/gnu/make/make-4.3.tar.gz && tar -xf make-4.3.tar.gz
(cd make-4.3 && ./configure && make)

wget -nv https://ftp.gnu.org/gnu/make/make-4.4.tar.gz && tar -xf make-4.4.tar.gz
(cd make-4.4 && ./configure && make)

# Try to use the Makefiles.

make-4.2/make -f Makefile
# "netgo,ledger"
make-4.3/make -f Makefile
# "netgo ledger,"
make-4.4/make -f Makefile
# "netgo ledger,"

make-4.2/make -f Makefile.proposal
# "netgo,ledger"
make-4.3/make -f Makefile.proposal
# "netgo,ledger"
make-4.4/make -f Makefile.proposal
# "netgo,ledger"

cd
rm -rf "$WORKSPACE"

---

For Admin Use

• Not duplicate issue
• Appropriate labels applied
• Appropriate contributors tagged
• Contributor assigned/self-assigned
• Is a spike necessary to map out how the issue should be approached?

[sigv]

It looks like Gaia v7.1.0 was built with Make 4.3+ as well:

$ ./gaiad-v7.1.0-linux-amd64 version --long 2>&1 | head -n5
name: gaia
server_name: gaiad
version: v7.1.0
commit: 5db8fcc9a229730f5115bed82d0f85b6db7184b4
build_tags: netgo ledger,

[yaruwangway]

Hi @sigv, thank you for opening this issue, can you check if this PR fixes what you addressed in this issue ? #2051

[yaruwangway]

ha, sry, just saw you already opened pr, will review! @sigv

[sigv]

ha, sry, just saw you already opened pr, will review!

No worries! And thank you!

Some background:

We (kjnodes) discovered this with lavanet/[email protected], where the -X flag was not quoted, resulting in build failure. Most derivate Cosmos-based chains appear to have copy-pasted the -X flag quoted, as visible in cosmos/[email protected]. This enables spaces in the $(build_tags_comma_sep) to build properly. From my understanding, this doesn't directly affect any operational aspect of the node itself, therefore backporting is entirely optional.

As for the proposed change itself, we proposed it to Lava Network and it was released in lavanet/[email protected]. It gets the job done. The test scenario to do is just spin up a fresh VM on 20.04 and 22.04, run builds with and without patch, compare output of version subcommand from the built binary. Looks reasonable to me.

Let me know if I can help with anything else!

[sigv]

@yaruwangway, any chance of having the change reviewed?

[yaruwangway]

Hi @sigv , i reviewed, all good, but we need two reviewers to get merged, i will ask my colleagues.

[sigv]

Much appreciated! 🙏🏻🥂