depends: Force `CMAKE_EXPORT_NO_PACKAGE_REGISTRY=TRUE` #32943

pull hebasto wants to merge 1 commits into bitcoin:master from hebasto:250710-package-registry changing 1 files +1 −0

hebasto commented at 7:20 PM on July 10, 2025: member

When using CMake policies 3.14 and below, the export(PACKAGE) command by default populates the user package registry, which is stored outside the build tree. Setting the CMAKE_EXPORT_NO_PACKAGE_REGISTRY variable disables this side effect.

In CMake 3.15 and later, this behavior is disabled by default, and the variable has no effect.

This PR forces CMAKE_EXPORT_NO_PACKAGE_REGISTRY=TRUE globally, rather than managing it for each dependency package individually rather. It may be reverted once all CMake-based packages have updated to policies 3.15 or newer.

Fixes #32938.

depends: Force `CMAKE_EXPORT_NO_PACKAGE_REGISTRY=TRUE`

When using CMake policies 3.14 and below, the `export(PACKAGE)` command
by default populates the user package registry, which is stored outside
the build tree. Setting the `CMAKE_EXPORT_NO_PACKAGE_REGISTRY` variable
disables this side effect.

In CMake 3.15 and later, this behavior is disabled by default, and the
variable has no effect.

44f3bae300

hebasto added the label Build system on Jul 10, 2025
DrahtBot commented at 7:20 PM on July 10, 2025: contributor



The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.



Code Coverage & Benchmarks

For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/32943.



Reviews

See the guideline for information on the review process.

Type Reviewers

ACK fanquake

Concept ACK theuni

If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.
theuni commented at 9:54 PM on July 10, 2025: member
Yikes. Concept ACK.
The export(PACKAGE) command does not populate the user package registry when CMP0090 is set to NEW unless the CMAKE_EXPORT_PACKAGE_REGISTRY variable explicitly enables it. When CMP0090 is not set to NEW then export(PACKAGE) populates the user package registry unless the CMAKE_EXPORT_NO_PACKAGE_REGISTRY variable explicitly disables it.

CMAKE_FIND_USE_PACKAGE_REGISTRY disables the User Package Registry in all the find_package() calls when set to FALSE.
Could we also set CMAKE_FIND_USE_PACKAGE_REGISTRY to false in the toolchain file as a belt-and-suspenders in case some future package manually opts-in using CMAKE_EXPORT_PACKAGE_REGISTRY ?
fanquake commented at 9:32 AM on July 11, 2025: member

This is going to need backporting, to fix the 29.x branch.
fanquake added the label Needs backport (29.x) on Jul 11, 2025
maflcko added the label DrahtBot Guix build requested on Jul 11, 2025
hebasto commented at 10:34 AM on July 11, 2025: member

Could we also set CMAKE_FIND_USE_PACKAGE_REGISTRY to false in the toolchain file as a belt-and-suspenders in case some future package manually opts-in using CMAKE_EXPORT_PACKAGE_REGISTRY ?

I don't think it prevents issues like #32938, which occur when the toolchain file is not used.
hebasto commented at 10:43 AM on July 11, 2025: member

This is going to need backporting, to fix the 29.x branch.

I think it should be also backported to the 28.x branch, considering https://github.com/bitcoin/bitcoin/commit/f59e9057e2aa596b54cf9e85bab35c3ead137547.

Additionally, we might want to advise users to delete the ~/.cmake/packages/libevent directory on their machines.

Type	Reviewers
ACK	fanquake
Concept ACK	theuni

DrahtBot commented at 11:39 PM on July 11, 2025: contributor

Guix builds (on x86_64) [untrusted test-only build, possibly unsafe, not for production use]

File	commit 12fb00fd4228ad5158fd4893e7804d5667ea700b<br>(master)	commit ddeac5ea78d921fdbe944f845298d789dacd80e5<br>(pull/32943/merge)
*-aarch64-linux-gnu-debug.tar.gz	`3e6db6b2a92ec38d...`	`f10dc8611163571e...`
*-aarch64-linux-gnu.tar.gz	`d237464c3da00c64...`	`3adb01adb4e0daba...`
*-arm-linux-gnueabihf-debug.tar.gz	`2fd2488b8c89516b...`	`fd9efe0cb8622124...`
*-arm-linux-gnueabihf.tar.gz	`4f1368bc070632a5...`	`450407434dddfaff...`
*-arm64-apple-darwin-codesigning.tar.gz	`1357f810bbf948c2...`	`e63994dad3aaa149...`
*-arm64-apple-darwin-unsigned.tar.gz	`81a345df3778b7d6...`	`9b0e64c9e5279611...`
*-arm64-apple-darwin-unsigned.zip	`5869393d614baec8...`	`5d270e31879d91e4...`
*-powerpc64-linux-gnu-debug.tar.gz	`a2b3a082c7cb48a2...`	`7dc062b22302d731...`
*-powerpc64-linux-gnu.tar.gz	`46f6a45c63a955a1...`	`6fcf250179c7fc59...`
*-riscv64-linux-gnu-debug.tar.gz	`6ed452da26e7ca4e...`	`46af5c01c14f166e...`
*-riscv64-linux-gnu.tar.gz	`322180311cdfb956...`	`6bcf62db61328cc0...`
*-x86_64-apple-darwin-codesigning.tar.gz	`471d1c93c5a2e400...`	`4c47cae003e6615e...`
*-x86_64-apple-darwin-unsigned.tar.gz	`03d0c949fe2c90f9...`	`e6c43b88ed478aac...`
*-x86_64-apple-darwin-unsigned.zip	`5aff0350ee7085ac...`	`80a12b155ffd5d58...`
*-x86_64-linux-gnu-debug.tar.gz	`880aa77d59900eab...`	`cc3979bfa15dad68...`
*-x86_64-linux-gnu.tar.gz	`73851d32a2ceafda...`	`50c1a4754e2b28a0...`
*.tar.gz	`85ca28c280c2db86...`	`abea6034ca07f068...`
SHA256SUMS.part	`9e2b78d3b518fd39...`	`82daf28609baaed1...`
guix_build.log	`46557ec7a496556c...`	`a3f6b4c4bac8d435...`
guix_build.log.diff		`66d38f9845fb6d98...`

DrahtBot removed the label DrahtBot Guix build requested on Jul 11, 2025
fanquake commented at 12:31 PM on July 14, 2025: member

ACK 44f3bae300dcafbe53f9b07e6cc22a112833e579

Yikes. Concept ACK.

Yea. It's an interesting example of how bumping/changing a CMake minimum version, can opt you into, or out of, unexpected behaviour (I imagine this is why this stopped being the default).
DrahtBot requested review from theuni on Jul 14, 2025
fanquake merged this on Jul 14, 2025
fanquake closed this on Jul 14, 2025
fanquake referenced this in commit bc2147c884 on Jul 14, 2025
fanquake commented at 12:35 PM on July 14, 2025: member

Backported to 29.x in #32863.
fanquake removed the label Needs backport (29.x) on Jul 14, 2025
fanquake referenced this in commit 792ee20318 on Jul 14, 2025
fanquake commented at 12:41 PM on July 14, 2025: member

Backported to 28.x in #32969.
hebasto deleted the branch on Jul 14, 2025
luke-jr referenced this in commit a19f529ab0 on Jul 17, 2025
glozow referenced this in commit ef380a454c on Jul 18, 2025
fanquake referenced this in commit a6d0159518 on Jul 23, 2025
sedited referenced this in commit 02ded863ba on Jul 28, 2025
sedited referenced this in commit 52d7f32bd6 on Jul 28, 2025
alexanderwiederin referenced this in commit 28fe919bf7 on Aug 6, 2025
sedited referenced this in commit b98d982d79 on Aug 7, 2025
alexanderwiederin referenced this in commit 4152176d02 on Aug 8, 2025
alexanderwiederin referenced this in commit 9ef94c31db on Aug 8, 2025
stringintech referenced this in commit 71275a1b5e on Aug 17, 2025
yuvicc referenced this in commit 22f55cf11d on Aug 26, 2025
ajtowns referenced this in commit a3b68db613 on Sep 4, 2025
bug-castercv502 referenced this in commit 4aa5572aea on Sep 28, 2025
stickies-v referenced this in commit a19c56cd7c on Nov 4, 2025
morozow referenced this in commit c127731ed1 on May 8, 2026
morozow referenced this in commit 6c4eaecbfa on May 8, 2026
morozow referenced this in commit d365167457 on May 8, 2026
morozow referenced this in commit 8214c3bc7c on May 8, 2026
Kino1994 referenced this in commit 8b9078f03f on Jun 28, 2026
BigcoinBGC referenced this in commit 703cdb95b7 on Jun 30, 2026
bitcoin locked this on Jul 30, 2026