Repoquery: Add --duplicates option #178
Conversation
kontura
force-pushed
the
repoquery_duplicates
branch
from
23d35e7
to
6657370
Compare
libdnf/rpm/package_query.cpp
Outdated
| for (int pos_second = start_block; pos_second < stop_block; ++pos_second) { | ||
| Id id_second = samename[pos_second]; | ||
| s_second = pool.id2solvable(id_second); | ||
| if ((s_first->evr == s_second->evr) && (s_first->arch != s_second->arch)) { |
There was a problem hiding this comment.
All packages in the block should have the same arch. Why is && (s_first->arch != s_second->arch) needed?
There was a problem hiding this comment.
It is not, I forgot that the arch check was moved. 👍 👍
Since thanks to your other suggestion we no longer compare each solvable to it self I think we can drop the condition entirely. If there somehow was a package installed twice with the same name, arch and even evr I think we would want it in the query. Packages with the same nevra but different id surely are duplicates in general for the filter_duplices().
What do you think?
libdnf/rpm/package_query.cpp
Outdated
| for (int pos_first = start_block; pos_first < stop_block; ++pos_first) { | ||
| Id id_first = samename[pos_first]; | ||
| s_first = pool.id2solvable(id_first); | ||
| for (int pos_second = start_block; pos_second < stop_block; ++pos_second) { |
There was a problem hiding this comment.
You compare each couple of solvables twice (i.e. pkgA with pkgB and than pkgB with pkgA again). I think you can start the inner cycle with for (int pos_second = pos_first.... Or even pos_second = pos_first + 1 to avoid comparison of solvable with itself.
kontura
force-pushed
the
repoquery_duplicates
branch
2 times, most recently
from
81ae48c
to
9ef25f8
Compare
kontura
force-pushed
the
repoquery_duplicates
branch
from
9ef25f8
to
ac6b4d3
Compare
Adds
filter_duplicates()method toPackageQuery.For: #122
Tests: rpm-software-management/ci-dnf-stack#1187