diff options
author | Nagy Gabor <ngaba@bibl.u-szeged.hu> | 2009-01-10 16:25:03 +0100 |
---|---|---|
committer | Dan McGee <dan@archlinux.org> | 2009-01-13 06:59:37 +0100 |
commit | cb7337eb640b70b295edb7b05b016b53352d8713 (patch) | |
tree | 29c28513a9ed07668902a06988cd9a1d74f96efc /pactest/README | |
parent | 4da70d800a25881f525498a451ae21b398a1d5b1 (diff) | |
download | pacman-cb7337eb640b70b295edb7b05b016b53352d8713.tar.gz pacman-cb7337eb640b70b295edb7b05b016b53352d8713.tar.xz |
db->pkgcache_loaded and db->grpcache_loaded
Clearly the old code was more elegant (NULL cache indicated "not loaded"),
but it had some drawbacks, so from now on we indicate the state of caches
explicitly.
Old drawbacks:
When we had an empty database (unstable), libalpm called db_populate after
every pkgcache access, because NULL pkgcache indicated "not loaded" state.
This is not a common case, but the same situation can happen with grpcache,
which is more problematic: If the user had a custom repo with no groups,
grpcache was always NULL. (grpcache is also loaded per database.) Thus
every get_grpcache call induced a load_grpcache operation, so the benefits
of grpcache was completely lost.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Dan McGee <dan@archlinux.org>
Diffstat (limited to 'pactest/README')
0 files changed, 0 insertions, 0 deletions