Hi Allan,
I will wait for the conversation to begin before diving into the deep discussions.
Just quickly ...
I am open to the language as long as STIX 2.x implementations do not raise
errors when they encounter SEP contents that they do not understand.
We have given up on using Custom Objects and Properties due to this very concern.
I am afraid that this concern may hamper SEP's idea of the process of gradually
moving custom objects/properties into the standard.
STIXPreferred is for an org that produces or consumes custom objects/properties and
it does not cover other orgs that happen to receive custom objects/properties.
Regards,
Ryu
From: cti@lists.oasis-open.org <cti@lists.oasis-open.org>
On Behalf Of aa tt
Sent: Saturday, August 22, 2020 6:00 AM
To: Masuoka, Ryusuke/çå
çä <masuoka.ryusuke@fujitsu.com>
Cc: Paul Patrick <ppatrick@darklight.ai>; cti@lists.oasis-open.org
Subject: Re: [cti] Updated Extension Proposal
Hi Ryu - Weâre still waiting on the chairs/co-chairs to convene a specific TC working call to discuss the proposal.
As soon as that call is scheduled we should include your points in that discussion.
However, I suggest that any language in the spec on things a consumer of STIX does not understand should not state âMUST processâ. The definition of âprocess' is ambiguous and therefore not testable without clarification.
I would suggest we consider carefully the language used for compliance very carefully so that test specification may be written and easily checked against. This will become especially important for STIXPreferred if/when that gets updated for STIX2.1.
If you take a look at how STIXPreferred currently expresses the interoperability of solutions handling custom properties/objects that might help guide any future changes to the spec when we are talking about handling
content defined as an extension.
Thank you for your input and sorry for my late response.
(I am back from my long summer (Obon) break.)
I too am supportive of replacing the existing customization mechanism with this proposal
and deprecating the current approach until a future release based on the following reasons.
- There should be only one way to do a certain thing.
I understand this is one of the lessons the TC learned from STIX 1.x experiences.
- The current approach ("11 Customizing STIX" of STIX 2.1) says
A consumer that receives STIX content containing Custom Properties, Objects or Extensions
it does not understand MAY refuse to process the
content or MAY ignore those properties
or objects and continue processing the content.
and some STIX 2.x implementation actually raises errors when they see
Custom Properties/Objects that it does not understand.
When this happens, it is really difficult to use Custom Properties/Objects and
it limits their utility/practicality. They can ignore what they do not understand, but
I would like the implementations to process the STIX as long as it is conformant.
I hope that SEP is going to be like "A consumer MAY ignore
the STIX it does not understand, but MUST process the STIX as long as
it is conformant (to SEP)." Am I right to expect that?
I generally agree with Allan in that I see the STIX Extension proposal as a candidate to replace the way customization of STIX 2.1 is done today. Like Allan,
I too would be supportive of replacing the existing customization mechanism with this proposal but would recommend that the TC deprecate, not remove, the current approach until a future release.
To that end, I believe the section in the spec on Customizing STIX be updated to not only provide guidance on how to use the STIX Extension approach but provide
guidance how to use it IN PLACE OF the existing schema. I can also imagine what an update to STIX Elevator could be to aid in stepping-up from STIX 2.1 to STIX 2.1.1 to make that transition easier for people to move to the new scheme.
My personal feeling is that customization of STIX will remain in the standard as-is but industry best-practices will steer orgs towards using the extension proposal as the primary mechanism to change STIX.
The rationale: JSON content can always be modified regardless of what the CTI TC decides and the customization language in the current specification can just remain as most people already have done customization in products.
However, if the TC felt that it was better to remove the customization language and replace with this extension proposal then I would be supportive of that also.
Other proponents of the extension proposal, as well as other TC members should speak up on this topic.
I might have missed some discussions, but
what are the relationships between
"11 Customizing STIX" of STIX 2.1 (custom properties,
custom objects, and custom extensions) and
Please find attached an updated STIX extension proposal that the mini-group worked with IBM (Jason/Emily) to incorporate an acceptable compromise to their concerns.
1) Previously presented extension proposal is maintained and has been named âsub-component extensionâ to
clarify the distinction between when it would/should be used vs the additional mechanism added for top-level extension.
2) Added the concept of Top-Level object extension to allow extension properties directly with existing STIX standard objects.
3) Clarified language/definitions around use for both aspects.
The google doc containing the specification changes are here:
This proposal represents the agreement by all mini-group participants. We would like to encourage the TC to consider this proposal and incorporate the changes to the STIX standard.
|