Created attachment 675022 [details] emerge --info output Portage is trying to update the rust compiler and have the system-bootstrap USE flag set, but I'm greeted with the following error: * Error: circular dependencies (dev-lang/rust-1.48.0:stable/1.48::gentoo, ebuild scheduled for merge) depends on (dev-lang/rust-1.48.0:stable/1.48::gentoo, ebuild scheduled for merge) (buildtime) I have dev-lang/rust-1.47.0-r1 installed, but if I'm understanding the ebuild for 1.48.0 correctly, it should only depend on 1.47: # to bootstrap we need at least exactly previous version, or same. # most of the time previous versions fail to bootstrap with newer # for example 1.47.x, requires at least 1.46.x, 1.47.x is ok, # but it fails to bootstrap with 1.48.x # https://github.com/rust-lang/rust/blob/${PV}/src/stage0.txt BOOTSTRAP_DEPEND="|| ( =dev-lang/rust-$(ver_cut 1).$(($(ver_cut 2) - 1))* =dev-lang/rust-bin-$(ver_cut 1).$(($(ver_cut 2) - 1))* =dev-lang/rust-$(ver_cut 1).$(ver_cut 2)* =dev-lang/rust-bin-$(ver_cut 1).$(ver_cut 2)* ) " BDEPEND="${PYTHON_DEPS} app-eselect/eselect-rust || ( >=sys-devel/gcc-4.7 >=sys-devel/clang-3.5 ) system-bootstrap? ( ${BOOTSTRAP_DEPEND} ) !system-llvm? ( dev-util/cmake dev-util/ninja ) "
can you please send exact accept_keywords for rust you have. (including virtual if any) exact command you run and attach output of your 'emerge --<yourargs> -pvd'
I saw weird behavior related to this as well. When I upgraded from rust 1.47.0-r1 to 1.48.0, rust-bin-1.48.0 was pulled in to bootstrap rust-1.48.0. I'm not sure why that was necessary since I already had rust 1.47.0-r1 on my system.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=e61ae40fcdeac8948f307c4dff57337e5d79caef commit e61ae40fcdeac8948f307c4dff57337e5d79caef Author: Georgy Yakovlev <gyakovlev@gentoo.org> AuthorDate: 2020-11-26 19:18:46 +0000 Commit: Georgy Yakovlev <gyakovlev@gentoo.org> CommitDate: 2020-11-26 19:19:32 +0000 dev-lang/rust: don't default to system-bootstrap for now Bug: https://bugs.gentoo.org/689336 Bug: https://bugs.gentoo.org/756793 Package-Manager: Portage-3.0.10, Repoman-3.0.2 Signed-off-by: Georgy Yakovlev <gyakovlev@gentoo.org> dev-lang/rust/rust-1.47.0-r1.ebuild | 2 +- dev-lang/rust/rust-1.47.0-r2.ebuild | 2 +- dev-lang/rust/rust-1.48.0.ebuild | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-)
yeah something strange is going on. on stable system I have rust-1.46.0 installed accepted keywords for =dev-lang/rust-1.47* =virtual/rust-1.47* and portage tries to pull rust-bin-1.46.0 for installation... very strange. disabled by default for now.
Created attachment 675193 [details] emerge dev-lang/rust virtual/rust -pvud > emerge-rust-bin-debug.txt 2>&1 another weirdness stable system has: dev-lang/rust-1.46.0 virtual/rust-1.46.0 accept_keywords: =dev-lang/rust-1.47* =virtual/rust-1.47* trying to emerge it: # emerge dev-lang/rust virtual/rust -pvu These are the packages that would be merged, in order: Calculating dependencies .... done! [ebuild N ] dev-lang/rust-bin-1.46.0:stable::gentoo USE="-clippy -doc -rls -rustfmt" ABI_X86="(64) -32 (-x32)" CPU_FLAGS_X86="sse2" 0 KiB [ebuild U ~] dev-lang/rust-1.47.0-r2:stable/1.47::gentoo [1.46.0:stable/1.46::gentoo] USE="clippy rustfmt system-bootstrap -debug (-doc) -libressl -miri -nightly -parallel-compiler -rls -system-llvm -test% -wasm" ABI_X86="(64) -32 (-x32)" CPU_FLAGS_X86="sse2" LLVM_TARGETS="(X86) -AArch64 -AMDGPU -ARM -AVR% -BPF -Hexagon -Lanai -MSP430 -Mips -NVPTX -PowerPC -RISCV -Sparc -SystemZ -WebAssembly -XCore" 0 KiB [ebuild U ~] virtual/rust-1.47.0::gentoo [1.46.0::gentoo] ABI_X86="(64) -32 (-x32)" 0 KiB Total: 3 packages (2 upgrades, 1 new), Size of downloads: 0 KiB so for some reason it pulls rust-bin-1.46.0, despite having rust-1.46.0 installed and available. output of > emerge dev-lang/rust virtual/rust -pvud > emerge-rust-bin-debug.txt 2>&1 attached.
depstring we are dealing here is > || > ( > =dev-lang/rust-1.46* > =dev-lang/rust-bin-1.46* > =dev-lang/rust-1.47* > =dev-lang/rust-bin-1.47* > )
(In reply to Georgy Yakovlev from comment #5) > Created attachment 675193 [details] > emerge dev-lang/rust virtual/rust -pvud > emerge-rust-bin-debug.txt 2>&1 > > another weirdness I've opened bug 756961 for this, and it looks like a misbehavior of this change related to bug 705986: https://gitweb.gentoo.org/proj/portage.git/commit/?id=ef66aab178180288132177011ad0e910fb2cd55e
(In reply to Georgy Yakovlev from comment #1) > can you please send exact accept_keywords for rust you have. (including > virtual if any) > exact command you run > > > and attach output of your 'emerge --<yourargs> -pvd' Sure. The accept_keywords I have are virtual/rust ~amd64 dev-lang/rust ~amd64 The exact command I ran was (expanded out through an function in my .bashrc and my EMERGE_DEFAULT_OPTS) emerge --ask --verbose --ask-enter-invalid --update --deep --with-bdeps=y @world
Created attachment 675241 [details] emerge debug output Output of my emerge command with -pd appended
(In reply to Oskari Pirhonen from comment #9) > Created attachment 675241 [details] > emerge debug output > > Output of my emerge command with -pd appended This looks like a similar problem to the log attached in comment #5. If you have masked dev-lang/rust-bin, then due to the issue described in bug 756961 it will fail with a circular dependency error and this suggestion: It might be possible to break this cycle by applying the following change: - dev-lang/rust-1.48.0 (Change USE: -system-bootstrap)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage.git/commit/?id=b991f23ad915d1fc37088f11ca855c8c8374f5ae commit b991f23ad915d1fc37088f11ca855c8c8374f5ae Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2020-11-28 01:57:14 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2020-11-28 20:50:13 +0000 Allow a package to replace its own buildtime dependency If a package has a buildtime dependency on a previous version that it will replace, then do not treat it as a slot conflict. This solves inappropriate behavior for dev-lang/rust[system-bootstrap]. This requires adjustments to package selection logic in several locations, in order to ensure that an installed package instance will be selected to satisfy a buildtime dependency when appropriate. Dependencies of the installed package will be entirely ignored, but that has already been the case when using installed package to break cycles, as discussed in bug 199856. Bug: https://bugs.gentoo.org/756793 Bug: https://bugs.gentoo.org/756961 Signed-off-by: Zac Medico <zmedico@gentoo.org> lib/_emerge/depgraph.py | 68 +++++++++++++--- lib/portage/dep/dep_check.py | 24 +++--- .../tests/resolver/test_circular_choices_rust.py | 94 ++++++++++++++++++++++ 3 files changed, 164 insertions(+), 22 deletions(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=d77f07243dac0d12e44d0b0c5e6ab79c76111809 commit d77f07243dac0d12e44d0b0c5e6ab79c76111809 Author: Zac Medico <zmedico@gentoo.org> AuthorDate: 2020-11-28 21:10:27 +0000 Commit: Zac Medico <zmedico@gentoo.org> CommitDate: 2020-11-28 21:17:14 +0000 sys-apps/portage: Bump to version 3.0.11 #756961 handle dev-lang/rust[system-bootstrap] dependency cycle #757306 backtracking: fix virtual choices for circular deps Bug: https://bugs.gentoo.org/756793 Bug: https://bugs.gentoo.org/756961 Bug: https://bugs.gentoo.org/757306 Bug: https://bugs.gentoo.org/756028 Package-Manager: Portage-3.0.11, Repoman-3.0.2 Signed-off-by: Zac Medico <zmedico@gentoo.org> sys-apps/portage/Manifest | 1 + sys-apps/portage/portage-3.0.11.ebuild | 267 +++++++++++++++++++++++++++++++++ 2 files changed, 268 insertions(+)
Thanks Zac! I confirm it behaves fine with portage-3.0.11 now to re-enable system-bootstrap I'll have to wait a bit before it goes stable and sits in the tree for a while.
Yup, I pulled in portage 3.0.11 and I can also confirm that it worked. The pre-existing rust 1.47 was used to build rust 1.48 with USE=system-bootstrap enabled. Which is great, because that lets me use "nightly" with -Zthreads=16 when building the compiler. Thanks a bunch Zac!
