for glibc: INSTALL_MASK=/usr/share/i18n/charmaps qmerge report /usr/share/i18n/charmaps as INSTALL_MASK INSTALL_MASK=/usr/share/i18n/charmaps -/usr/share/i18n/charmaps/UTF-8.gz This cancels out both, no INSTALL_MASK is reported by qmerge INSTALL_MASK=-/usr/share/i18n/charmaps/UTF-8.gz /usr/share/i18n/charmaps qmerge reports only /usr/share/i18n/charmaps as INSTALL_MASK The INSTALL_MASK code is a bit complicated for me to make sense of so any ideas appreciated
I remember negations can only be applied after the mask it works on is known, so the 3rd case is correct. 1st case too, the 2nd is unexpected, because this was the testing case of that code.
Not sure case 2 is invalid see: https://archives.gentoo.org/gentoo-portage-dev/message/7c6e5ec23d9af9ee1f19dd322f2c2e53 : That won't work since exclusive and inclusive patterns can be stacked, and for this to work they are applied in order. E.g. /usr/foo -/usr/foo/bar /usr/foo/bar/baz Should remove all foo subdirectories except for bar, and also baz inside bar.
case 2 is valid, and it's wrong that the entire mask is cancelled.
did you by chance compile the code using -DEBUG=1? This should make install_mask_pwd output a blurb with computed state.
Ok, I can reproduce, this is messed up, I'm sure it worked (but I guess I should've added a test...)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage-utils.git/commit/?id=61fdfd4d8cc40da77455c6f8b36a3ec36aa5b146 commit 61fdfd4d8cc40da77455c6f8b36a3ec36aa5b146 Author: Fabian Groffen <grobian@gentoo.org> AuthorDate: 2020-01-03 11:26:49 +0000 Commit: Fabian Groffen <grobian@gentoo.org> CommitDate: 2020-01-03 11:26:49 +0000 tests/qmerge: add case for INSTALL_MASK This tests INSTALL_MASK on a simple and a inversed path, with one level difference. Bug: https://bugs.gentoo.org/699376 Signed-off-by: Fabian Groffen <grobian@gentoo.org> tests/qmerge/dotest | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+)
I have a feeling the multi-level thing is the problem.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage-utils.git/commit/?id=fbf7ee544f0efeb6f5a8254f6a994d1daa5a4573 commit fbf7ee544f0efeb6f5a8254f6a994d1daa5a4573 Author: Fabian Groffen <grobian@gentoo.org> AuthorDate: 2020-01-03 12:39:02 +0000 Commit: Fabian Groffen <grobian@gentoo.org> CommitDate: 2020-01-03 12:39:02 +0000 tests/qmerge: add 2-level negation INSTALL_MASK test Bug: https://bugs.gentoo.org/699376 Signed-off-by: Fabian Groffen <grobian@gentoo.org> TODO.md | 3 +++ qmerge.c | 8 ++++---- tests/qmerge/dotest | 15 ++++++++++++++- tests/qmerge/packages/Packages | 6 +++--- tests/qmerge/packages/sys-devel/qmerge-test-1.3.tbz2 | Bin 7264 -> 7289 bytes 5 files changed, 24 insertions(+), 8 deletions(-)
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/proj/portage-utils.git/commit/?id=4d6b231cf0973e1d3b869d98a0e263155f04b20c commit 4d6b231cf0973e1d3b869d98a0e263155f04b20c Author: Fabian Groffen <grobian@gentoo.org> AuthorDate: 2020-01-03 12:43:49 +0000 Commit: Fabian Groffen <grobian@gentoo.org> CommitDate: 2020-01-03 12:43:49 +0000 tests/qmerge: test masking an intermediate level in INSTALL_MASK this tests the case from the archives: /usr/foo -/usr/foo/bar /usr/foo/bar/baz Bug: https://bugs.gentoo.org/699376 Signed-off-by: Fabian Groffen <grobian@gentoo.org> tests/qmerge/dotest | 12 ++++++++++++ 1 file changed, 12 insertions(+)
all tests seem conclusive (that it works) Need to narrow down a case that is wrong.
Can we get an error scenario for 0.84? I've added some tests which don't seem to indicate any problem for the scenario's you've listed.
