OWASP Common Numbering Project
Below is the latest proposal for your consideration. Questions/Comments? Email Mike.
OWASP-06 OWASP-06-DEPRECATED OWASP-0604 OWASP-0604-DEPRECATED OWASP-0604-DG OWASP-0604-DG-01 OWASP-0604-TG OWASP-0604-TG-DV-005 OWASP-0604-TG-DV-005-DEPRECATED
0123456789012345678901234567890123456789 1 2 3
- 0-4 OWASP
- 6-7 Detailed requirement identifier (major)
- 8-9 Detailed requirement identifier (minor)
- 11-12 Document code (DG=Development Guide, TG=Testing Guide, CG=Code Review Guide, AR, ED, RM, OR, others reserved)
- 14-40 (Optional: DEPRECATED, or # for iterations, or legacy identifiers)
Notes and Guidance
- adding the (release) year into the numbering scheme can be problematic, because the document has a life cycle that goes over years ....
- One should rather try to accommodate a versioning scheme that is human readable in the reference number as well (e.g. V02, or RevA, or...)
- don't try to encode any information into the ID that is likely to change or be subject to debate. In the olden days of CVE, we used to have "CAN-1999-0067" which would change into "CVE-1999-0067" once the item was considered stable and sufficiently verified. That made the ID hard to use. Right now, OWASP-DV-001 encodes the term "data validation" in the DV acronym, but what happens if in a couple of years, some new and better term occurs, or the focus changes from validation to something else? (As an example, it's only recently that the "data validation" term itself has become popular.)
- carefully consider the range of values that your ID space supports, and if possible, allow it to expand. CVE has a "CVE-10K" problem because we never expected that we would ever come close to tracking 10,000 vulnerabilities a year. Red Hat had to change their advisory numbering scheme a couple years ago. etc.
- don't change the fundamental meaning of the ID once you've assigned it. This causes confusion, and more importantly, it immediately invalidates almost everyone's mappings to that ID - including people who you don't even know are using that ID.
- closely monitor the mappings that get made. Typos and misunderstandings are rarely caught. People may make assumptions about what "the item" really is, based only on a quick scan of a short name or title. Since you're dealing with diverse sources, there are likely to be many-to-many relationships in dealing with mappings.
- determine some kind of procedure for handling duplicates. They're gonna happen.
- the more you distribute the process of creating and assigning IDs between multiple people, the more inconsistencies and duplicates you will wind up with. This may be unavoidable, since the job is usually bigger than one person.
- determine some kind of procedure for deprecating IDs, i.e., "retiring" them and discouraging their use by others. This will probably happen for reasons other than duplicates. There should be some final record, somewhere, of what happened to the deprecated item - i.e., it shouldn't just disappear off the face of the earth.
Much of the discussion surrounding the establishment of "Common OWASP Numbering" can be found on the various OWASP mailing lists. (For your convenience here is a direct link to the OWASP Testing Guide Mailing List Archive.)