You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Additional context:
Recently, while adding dependency relationships, I found these confusing to follow due in part to this discrepancy in naming. From is mapped to Parent, so adding a dependency relationship is: From: PKG-A Dependency-Of To: PKG-B, to indicate PKG-B has a dependency of PKG-A, or PKG-A is a dependency of PKG-B depending on how you prefer to phrase the relationship. However this maps to Parent: PKG-A and Child: PKG-B, which seems backwards in terms of dependencies: I would consider the "parent" in a dependency relationship to be the one containing the other.
Additionally, if we are modifying relationships, I would strongly encourage considering using a one-to-many relationship model, similar to SPDX 3. I believe this would help to significantly reduce the size of SBOMs but would mean that we would have to invert Dependency-Of and only allow for relationship types of similar directionality.
What would you like to be added:
Align the terminology used to define relationships.
Today, in the Syft internal data model, relationships have the fields:
From
andTo
, but the Syft JSON hasParent
andChild
. These are all part of the public Syft 1.0 API at this point.Additional context:
Recently, while adding dependency relationships, I found these confusing to follow due in part to this discrepancy in naming.
From
is mapped toParent
, so adding a dependency relationship is: From: PKG-A Dependency-Of To: PKG-B, to indicate PKG-B has a dependency of PKG-A, or PKG-A is a dependency of PKG-B depending on how you prefer to phrase the relationship. However this maps to Parent: PKG-A and Child: PKG-B, which seems backwards in terms of dependencies: I would consider the "parent" in a dependency relationship to be the one containing the other.Additionally, if we are modifying relationships, I would strongly encourage considering using a one-to-many relationship model, similar to SPDX 3. I believe this would help to significantly reduce the size of SBOMs but would mean that we would have to invert
Dependency-Of
and only allow for relationship types of similar directionality.Example from this pseudo-proposal:
The text was updated successfully, but these errors were encountered: