| Message ID | 20181017154046.12299-1-morganamilo@gmail.com |
|---|---|
| State | Accepted, archived |
| Delegated to: | Andrew Gregory |
| Headers | show |
On 10/17/18 at 04:40pm, morganamilo wrote: > When --needed is used, up to date packages are now filtered out > before showing the group select. > > Fixes FS#22870. > > Signed-off-by: morganamilo <morganamilo@gmail.com> > --- > v2: Changed per Andrew's feedback. I wasn't concerned about the misleading message for this patch, but the failing test is another thing. If you're not testing your patches with `make check`, please do. We could update the test, but let's just go ahead and fix the "error" after all. I recommend, in a separate patch, adding a group_exists function and using that to distinguish between a non-existent group and a group with no packages selected after find_group_packages returns NULL. > diff --git a/lib/libalpm/sync.c b/lib/libalpm/sync.c > index b6ae7b72..05f58fad 100644 > --- a/lib/libalpm/sync.c > +++ b/lib/libalpm/sync.c > @@ -277,10 +277,21 @@ alpm_list_t SYMEXPORT *alpm_find_group_pkgs(alpm_list_t *dbs, > > for(j = grp->packages; j; j = j->next) { > alpm_pkg_t *pkg = j->data; > + alpm_trans_t *trans = db->handle->trans; > > if(alpm_pkg_find(ignorelist, pkg->name)) { > continue; > } > + if(trans != NULL && trans->flags & ALPM_TRANS_FLAG_NEEDED) { > + alpm_pkg_t *local = _alpm_db_get_pkgfromcache(db->handle->db_local, pkg->name); > + if(local && _alpm_pkg_compare_versions(pkg, local) == 0) { > + /* with the NEEDED flag, packages up to date are not reinstalled */ > + _alpm_log(db->handle, ALPM_LOG_WARNING, _("%s-%s is up to date -- skipping\n"), > + local->name, local->version); > + ignorelist = alpm_list_add(ignorelist, pkg); > + continue; > + } > + } > if(alpm_pkg_should_ignore(db->handle, pkg)) { > alpm_question_install_ignorepkg_t question = { > .type = ALPM_QUESTION_INSTALL_IGNOREPKG, > -- > 2.19.1
On Thu, 18 Oct 2018 at 01:34, Andrew Gregory <andrew.gregory.8@gmail.com> wrote: > I wasn't concerned about the misleading message for this patch, but > the failing test is another thing. If you're not testing your patches > with `make check`, please do. We could update the test, but let's > just go ahead and fix the "error" after all. I recommend, in > a separate patch, adding a group_exists function and using that to > distinguish between a non-existent group and a group with no packages > selected after find_group_packages returns NULL. I did say the patch set was incomplete ;) group_exists does seem like the best idea though so will do. Would you prefer that function be added to the backend or frontend?
diff --git a/lib/libalpm/sync.c b/lib/libalpm/sync.c index b6ae7b72..05f58fad 100644 --- a/lib/libalpm/sync.c +++ b/lib/libalpm/sync.c @@ -277,10 +277,21 @@ alpm_list_t SYMEXPORT *alpm_find_group_pkgs(alpm_list_t *dbs, for(j = grp->packages; j; j = j->next) { alpm_pkg_t *pkg = j->data; + alpm_trans_t *trans = db->handle->trans; if(alpm_pkg_find(ignorelist, pkg->name)) { continue; } + if(trans != NULL && trans->flags & ALPM_TRANS_FLAG_NEEDED) { + alpm_pkg_t *local = _alpm_db_get_pkgfromcache(db->handle->db_local, pkg->name); + if(local && _alpm_pkg_compare_versions(pkg, local) == 0) { + /* with the NEEDED flag, packages up to date are not reinstalled */ + _alpm_log(db->handle, ALPM_LOG_WARNING, _("%s-%s is up to date -- skipping\n"), + local->name, local->version); + ignorelist = alpm_list_add(ignorelist, pkg); + continue; + } + } if(alpm_pkg_should_ignore(db->handle, pkg)) { alpm_question_install_ignorepkg_t question = { .type = ALPM_QUESTION_INSTALL_IGNOREPKG,
When --needed is used, up to date packages are now filtered out before showing the group select. Fixes FS#22870. Signed-off-by: morganamilo <morganamilo@gmail.com> --- v2: Changed per Andrew's feedback.