featuturefmt: Extract PotentialNamespace #722
Conversation
|
@KeyboardNerd The PR is ready to be reviewed. Here is the feature lister which uses a potential namespace which I used for testing. There is one thing which is causing problems now and I believe it will be resolved once you made your changes in updater.go When feature lister detects multiple namespaces (cpes) it returns each package multiple times with different CPE. It works fine, store namespaces, features, layer_feature, but it fails when storing data to ancestry_feature because of duplicated reference to namespaced_feature_id. |
|
I found one more bug, wait with a review, please. |
Fixed |
database/pgsql/layer.go
Outdated
| WHERE lf.feature_id = f.id | ||
| AND t.id = f.type | ||
| AND lf.namespace_id = ns.id |
There was a problem hiding this comment.
what if ls.namespace_id is null? there won't be a matched ns.id.
There was a problem hiding this comment.
Resolved by using Left join
|
Schema looks ok. Please add some tests on the change, especially the features with potential namespace. |
PotentialNamespace is feature namespace extracted while detecting features in layer. It will server for special feature detector. The current detectors return empty namespace.
PotentialNamespace is part of layer_feature table and it is also stored in namespace table.
PotentialNamespace should be in LayerFeature instead of Feature struct. Feature extractors were updated to return LayerFeature instead of Feature.
If layer contains more than one potential namespace, features will be created for each namespace. Layer_feature table now has to contains one more constrains (namespace_id).
Allda
force-pushed
the
feature_ns
branch
2 times, most recently
from
8a9db6e
to
16c9f37
Compare
| @@ -2,7 +2,8 @@ | |||
| INSERT INTO namespace (id, name, version_format) VALUES | |||
| (1, 'debian:7', 'dpkg'), | |||
| (2, 'debian:8', 'dpkg'), | |||
| (3, 'fake:1.0', 'rpm'); | |||
| (3, 'fake:1.0', 'rpm'), | |||
| (4, 'cpe:/o:redhat:enterprise_linux:7::server', 'rpm'); | |||
There was a problem hiding this comment.
Does 'cpe:/o:redhat:enterprise_linux:7::server' has any version information?
If this is the actual format for CPE, I'll have to change the namespace db a bit because right now we assume namespace name consists of two components: <name>:<version> as a string.
There was a problem hiding this comment.
Also, please rebase to master, and test. We found a bug in our tests that prevents database test failure from showing up.
There was a problem hiding this comment.
The version is usually included in CPE (in this case it is 7), but from the definition of CPE the version can be omitted.
There was a problem hiding this comment.
Yes, I noticed the bug. Today I spend a few hours trying to figure out why my test is not failing and after I found the cause and I fixed it I noticed that you fixed it yesterday
Test verifies that potential namespace is stored in database and it can be loaded back to structure. The commit also fixes few typos and bugs.
String was used when Feature contains PotentialNamespace. Since it was moved to LayerFeature we can use struct as map key instead of string.
PotentialNamespace is feature namespace extracted while detecting features in a layer. The PotentialNamespace is stored in nemaspace table + reference is also stored in layer_feature.
This will allow having multiple namespaces for single layer feature.
Partially resolves: #706