app/compose: include rpmdb pkglist in compose #1134
Conversation
src/libpriv/rpmostree-core.c
Outdated
| gv_nevra_from_pkg (DnfPackage *pkg) | ||
| { | ||
| return g_variant_ref_sink (g_variant_new ("(sstsss)", | ||
| dnf_package_get_nevra (pkg), |
There was a problem hiding this comment.
Refresh my memory why we have both the composed/decomposed versions of this?
There was a problem hiding this comment.
For the overrides case, it's really just to make it easier to consume the output of --json in tests. Since bandwidth is a consideration here, I think we can just reassemble it client-side. Will fix up!
| return FALSE; | ||
|
|
||
| GVariantBuilder pkglist_v_builder; | ||
| g_variant_builder_init (&pkglist_v_builder, (GVariantType*)"av"); |
There was a problem hiding this comment.
I think this should be a(stss) or so - the individual variants shouldn't vary.
jlebon
force-pushed
the
pr/rpmdb-in-commit-metadata
branch
from
87c55fa to
2f7e722
Compare
|
Fix up and test! ⬆️ |
|
LGTM. |
|
interesting. Does this have implications for #558 and ostreedev/ostree#954 ? |
Should make it much easier, yes. Though if we want to support history prior to commits that do not have this metadata, then it complicates things. We can discuss that in the issue! |
|
Let's see how the infra is today. |
We should be returning `FALSE` here, not `EXIT_FAILURE`. Closes: #1134 Approved by: cgwalters
Basically, it doesn't make sense for the caller to only want the pkglist, but not the refsack because the former has a more limited lifetime than the latter. Check for that to make sure nobody falls in this trap like I did. Closes: #1134 Approved by: cgwalters
We don't want to have to download all of `/usr/share/rpm` just to get the list of packages used to compose the tree. This is fundamental information that needs to be easier to discover. So let's stick it right in the commit metadata. There's various use cases for this information, including easily checking for and displaying updates and a pkglist-aware version of `ostree log`. Closes: #1134 Approved by: cgwalters
I'm not too worried about that. Onward and upward! |
|
💔 Test failed - status-atomicjenkins |
|
Oh heh, we don't actually have |
|
bot, retest this please |
|
🔒 Merge conflict |
We should be returning `FALSE` here, not `EXIT_FAILURE`.
Basically, it doesn't make sense for the caller to only want the pkglist, but not the refsack because the former has a more limited lifetime than the latter. Check for that to make sure nobody falls in this trap like I did.
We don't want to have to download all of `/usr/share/rpm` just to get the list of packages used to compose the tree. This is fundamental information that needs to be easier to discover. So let's stick it right in the commit metadata. There's various use cases for this information, including easily checking for and displaying updates and a pkglist-aware version of `ostree log`.
jlebon
force-pushed
the
pr/rpmdb-in-commit-metadata
branch
from
acee00c to
e11345d
Compare
|
Rebased! ⬆️ |
|
⚡ Test exempted: pull fully rebased and already tested. |
Basically, it doesn't make sense for the caller to only want the pkglist, but not the refsack because the former has a more limited lifetime than the latter. Check for that to make sure nobody falls in this trap like I did. Closes: #1134 Approved by: cgwalters
We don't want to have to download all of `/usr/share/rpm` just to get the list of packages used to compose the tree. This is fundamental information that needs to be easier to discover. So let's stick it right in the commit metadata. There's various use cases for this information, including easily checking for and displaying updates and a pkglist-aware version of `ostree log`. Closes: #1134 Approved by: cgwalters
We don't want to have to download all of
/usr/share/rpmjust to getthe list of packages used to compose the tree. This is fundamental
information that needs to be easier to discover. So let's stick it right
in the commit metadata. There's various use cases for this information,
including easily checking for and displaying updates and a pkglist-aware
version of
ostree log.