We noticed sys-apps/ed-1.19 was causing substantial slowdowns for world upgrades and emerging even ed-1.19 itself, but ed-1.18 is fine. ``` $ diff -ruN ed-1.{18,19} --- ed-1.18 2023-01-13 08:28:35.632365180 +0000 +++ ed-1.19 2023-01-13 08:29:28.822655700 +0000 @@ -1,13 +1,13 @@ -BDEPEND=sys-apps/texinfo || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) verify-sig? ( sec-keys/openpgp-keys-antoniodiazdiaz ) verify-sig? ( app-crypt/gnupg >=app-portage/gemato-16 ) +BDEPEND=sys-apps/texinfo || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) || ( >=app-arch/xz-utils-5.4.0 app-arch/plzip app-arch/pdlzip app-arch/lzip ) verify-sig? ( sec-keys/openpgp-keys-antoniodiazdiaz ) verify-sig? ( app-crypt/gnupg >=app-portage/gemato-16 ) DEFINED_PHASES=configure unpack DESCRIPTION=Your basic line editor -EAPI=7 +EAPI=8 HOMEPAGE=https://www.gnu.org/software/ed/ -INHERIT=toolchain-funcs unpacker verify-sig +INHERIT=edo toolchain-funcs unpacker verify-sig IUSE=verify-sig -KEYWORDS=~alpha amd64 arm arm64 hppa ~ia64 ~loong ~m68k ~mips ppc ppc64 ~riscv ~s390 sparc x86 ~x64-cygwin ~amd64-linux ~x86-linux ~ppc-macos ~x64-macos ~sparc-solaris ~sparc64-solaris ~x64-solaris ~x86-solaris +KEYWORDS=~alpha ~amd64 ~arm ~arm64 ~hppa ~ia64 ~loong ~m68k ~mips ~ppc ~ppc64 ~riscv ~s390 ~sparc ~x86 ~x64-cygwin ~amd64-linux ~x86-linux ~ppc-macos ~x64-macos ~sparc-solaris ~sparc64-solaris ~x64-solaris ~x86-solaris LICENSE=GPL-2+ SLOT=0 -SRC_URI=mirror://gnu/ed/ed-1.18.tar.lz verify-sig? ( mirror://gnu/ed/ed-1.18.tar.lz.sig ) -_eclasses_=toolchain-funcs ec8bbc28394c7305b43ea07164bd8f5f multilib 5ca4e49abed8e3a2f7b56920eadee157 multiprocessing b4e253ab22cef7b1085e9b67c7a3b730 unpacker 8bbea377a8dd2b5d5c8bc0d834ae6157 verify-sig afb98fe4696292e4234ce36713fd4e9e -_md5_=bf63ad1b4806a01a7000e3fe6ce90271 +SRC_URI=mirror://gnu/ed/ed-1.19.tar.lz https://download.savannah.gnu.org/releases/ed/ed-1.19.tar.lz verify-sig? ( mirror://gnu/ed/ed-1.19.tar.lz.sig https://download.savannah.gnu.org/releases/ed/ed-1.19.tar.lz.sig ) +_eclasses_=edo c0eb9cbe6b0bd01fcb4918f12598a4d3 toolchain-funcs ec8bbc28394c7305b43ea07164bd8f5f multilib 5ca4e49abed8e3a2f7b56920eadee157 multiprocessing b4e253ab22cef7b1085e9b67c7a3b730 unpacker 8bbea377a8dd2b5d5c8bc0d834ae6157 verify-sig afb98fe4696292e4234ce36713fd4e9e +_md5_=3f460bcb7087cbbadf886cad16e886c9 ``` unpacker.eclass's unpacker_src_uri_depends emits repeated || ( ... ) groups even ignoring USE flags. It should de-duplicate/flatten. Obviously there's a Portage bug here too in not merging the groups though.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=2575ec6caaa80ed8d3c9361312c797837c7c2ec3 commit 2575ec6caaa80ed8d3c9361312c797837c7c2ec3 Author: Sam James <sam@gentoo.org> AuthorDate: 2023-01-16 16:04:22 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2023-01-16 16:05:08 +0000 sys-apps/ed: workaround unpacker.eclass/Portage dep resolution bug 1. unpacker_src_uri_depends doesn't flatten the deps (bug #891133) and emits several || ( ... ) blocks (4). 2. Portage doesn't handle several repeated identical || ( ...) blocks correctly and takes ages to resolve (bug #891137). It should merge them together. Bug: https://bugs.gentoo.org/891137 Bug: https://bugs.gentoo.org/891133 Signed-off-by: Sam James <sam@gentoo.org> sys-apps/ed/{ed-1.19.ebuild => ed-1.19-r1.ebuild} | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)
It could use an assoc array rather than appending to be simple e.g. local -A deps *.zip) deps[zip]="app-arch/unzip" ;; *.lz) deps[lz]="|| ( ... )" ;; echo "${deps[*]}"
(In reply to Ionen Wolkens from comment #2) > It could use an assoc array rather than appending to be simple ...or if don't want to use hash order, could use numbers with a normal array instead I guess.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=ea08b1a398b28d8e03fee32787c70b227f763b02 commit ea08b1a398b28d8e03fee32787c70b227f763b02 Author: Sam James <sam@gentoo.org> AuthorDate: 2023-01-16 17:24:37 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2023-01-16 20:55:05 +0000 unpacker.eclass: flatten unpacker_src_uri_depends dependencies Populate an associative array as we iterate over SRC_URI to collect needed dependencies to avoid recording the same dependencies twice. This still doesn't handle USE flags, but it's significantly better than before, as we won't repeatedly emit the same dependency if there's more than one distfile in SRC_URI with the same suffix. Closes: https://bugs.gentoo.org/891133 Thanks-to: Ionen Wolkens <ionen@gentoo.org> Signed-off-by: Sam James <sam@gentoo.org> eclass/unpacker.eclass | 29 ++++++++++++++--------------- 1 file changed, 14 insertions(+), 15 deletions(-) Additionally, it has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=bf5cff8a6edd64fee318440582d3da3aa83a9043 commit bf5cff8a6edd64fee318440582d3da3aa83a9043 Author: Sam James <sam@gentoo.org> AuthorDate: 2023-01-16 17:26:50 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2023-01-16 20:55:05 +0000 sys-apps/ed: use unpacker_src_uri_depends again Now that unpacker_src_uri_depends is fixed to not emit duplicates, start using it again. (Note that unlike when we dropped it, no revbump needed to restore it, because the result should be the same now the eclass is fixed, rather than before where we were correcting the bad large deps.) Bug: https://bugs.gentoo.org/891133 Signed-off-by: Sam James <sam@gentoo.org> sys-apps/ed/ed-1.19-r1.ebuild | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)
