The CPE Goal: 1) Match CVEs to Releases 2) Match CVEs to Builds The CPE Problem: 1) CVEs * May be missing CPEs (e.g. all new CVEs) * May be missing YP matches (false negatives) * May have errors (false positives) 2) YP Recipes * No existing CPE map list * Recipe naming does not fully match package naming * Package version numbers may not exactly match Proposed CPE Solution: 1) Allow for SRT-based CPEs to fill in the matching gaps 2) Create a "Package" object that can map NIST-CPEs to YP-CPEs 3) Start maintaining a Recipe-To-Package table 4) Allow for flexible name/version mapping [any|version(s)], with confidence levels
We have been working on an implementation, and it will be ready for YP-3.3
It turns out that this is very very hard, give the poor mapping of CPEs to recipes, even when CPEs are avialable. My new strategy is to find ways to integrate with existing second and third tools in an agnostic manner.
Perhaps data from CVE Checker can help provide advice on how to proceed. Wind River's solution is that our triage management identifies the recipe from the CVE (mostly form the description given that the CPE table is usually empty for new CVEs), and our defect system tracks the CVEs by those recipe names. Ideally we have at least semi-automated assistance in those mappings. David
I am moving this away from "NEEDINFO". While this feature request definity needs more information about systemic difficulties in finding a resolution, it does not need the urgency of NEEDINFO.
This needs a mandate to implement. Will re-file new defect when that happens.