Stream: fmg
Topic: IGs vs THO
John Moehrke (Jan 19 2022 at 16:20):
I have heard of many IGs that have received a comment that defining URL for vocabulary should be anchored in THO. (see FHIR-35612 ). I don't think there is such a rule, and it does not seem appropriate as a general rule. It certainly is good QA modeling rules. But good QA modeling rules should not be appearing as member ballot comments.
Melva Peters (Jan 19 2022 at 17:50):
I believe there were about 250 comments in this last ballot cycle with this comment across all IGs
Lloyd McKenzie (Jan 20 2022 at 01:22):
The rule is that if an artifact is FMM3+, it's expected to leverage an external terminology (which includes THO). Continuing to use a FHIR-managed terminology for anything other than elements of type 'code' requires approval from the TSMG. (Which generally means that there's an agreement that it's not going to be useful anywhere else, that there's sufficient oversight of the terminology quality of the content, and that maintenance on an ever 2.5-3.5 year basis is sufficient.) Prior to FMM3, then local terminologies are fine, but they should have a URL that indicates that they're "temporary".
Lloyd McKenzie (Jan 20 2022 at 01:23):
However, for IGs, we're generally not surfacing FMM yet, so it might be a bit difficult to know what set of rules to apply. The general rule though is:
- url that indicates it's temporary
- external (e.g. THO) URL
- approval received from TSMG
Lloyd McKenzie (Jan 20 2022 at 01:24):
This will eventually be a QA rule for HL7-maintained IGs
John Moehrke (Jan 20 2022 at 12:25):
I can understand the intent, but I think this rule needs to be defined better than even what you express.
- I think it is quite common for a valueSet to be defined and only live in an IG, as the valueSet is specific to the use-cases covered by the valueSet. In some cases the valueSet is there purely to enable rules such as slicing.
- what does a temporary url look like? Have we ever defined this concept?
- TSMG needs to be far more approachable.
- I still can't submit anything, as I can't get the terminology build to succeed. This alone keeps me from addressing 5 CR for FHIR core.
- Not clear to me how one addresses a CR filed against FHIR core or an IG and re-invents that CR to submit to TSMG in a way that results in original CR getting addressed properly.
- the HL7 co-chairs hardly understand how to approach TSMG, and many IGs are written by non HL7 members
John Moehrke (Jan 20 2022 at 12:26):
More to my original point... There are now 250 comments against IGs that do not have the backing of a formal policy, rule, or procedure.
John Moehrke (Jan 20 2022 at 14:33):
note also that these 250 comments are backed by negative ballots.
Eric Haas (Jan 20 2022 at 15:30):
Specifically, I would like fmg to provide us with guidance and the language to tell the commenter to go pound sand (in a nice way) and dispose of all of these comments across IGs as not persuasive.
Lloyd McKenzie (Jan 20 2022 at 16:40):
FMG is unlikely to do that. The work group will need to do one of three things:
- change to a URL that indicates the code system is temporary
- put their terminology into an external source
- get TSMG approval to have their code systems maintained within their IG
John Moehrke (Jan 20 2022 at 16:41):
isn't that a rule that the IG must meet by FMM3+? Thus, if you are not yet there... what must one do?
John Moehrke (Jan 20 2022 at 16:41):
How is a URL indicated as temporary?
Lloyd McKenzie (Jan 20 2022 at 16:42):
If you're below FMM 3, the first bullet applies. And you can seek TSMG at any point. If you're FMM3 or above, you must have done one of the latter two. So regardless of FMM, one of the 3 bullets holds.
John Moehrke (Jan 20 2022 at 16:42):
is this just for codeSystems? Some of the comments I see are against ValueSets.
Lloyd McKenzie (Jan 20 2022 at 16:43):
For value sets, no. There has to be a good reason to put a value set into THO - i.e. it's both complex in its definition, it's likely to be re-used elsewhere, and the rules around its maintenance can be clearly expressed such that maintenance by a process other than WG ballot is 'safe'.
John Moehrke (Jan 20 2022 at 16:44):
If one gets THO exclusion for a CodeSystem, how is THAT indicated... where there is some unclear "temporary" flag, is there a flag that indicates "THO is okay with this being an IG specific code"?
Lloyd McKenzie (Jan 20 2022 at 16:44):
That will tend to be uncommon. So a "go pound sand" is a more likely disposition there.
John Moehrke (Jan 20 2022 at 16:44):
having governance enforced by ballot comments is not a proper solution
Lloyd McKenzie (Jan 20 2022 at 16:44):
Long term, it'll be a suppression of a warning in the QA with a reference to the TSMG minutes where they approved the exception.
Lloyd McKenzie (Jan 20 2022 at 16:44):
We just don't have the warning showing up in QA yet.
John Moehrke (Jan 20 2022 at 16:45):
I have one.... in an IG I am writing I have specialized List resource with a code for that kind of List, where that kind of List is only used in that IG.
Lloyd McKenzie (Jan 20 2022 at 16:45):
It's something that was agreed with vocab and HTA and FHIR-I and MnM a couple of cycles ago. And I'm pretty sure we've talked about it at FMG. We've just been slow to get it into the tools or communicate out to the work groups.
John Moehrke (Jan 20 2022 at 16:48):
Lloyd McKenzie said:
We just don't have the warning showing up in QA yet.
so, this is why we need a bulk action... lets get QA in effect, and bulk kill these CR that are demanding (negative ballot comments) that some governance that is not yet stated be retroactively applied.
John Moehrke (Jan 20 2022 at 16:49):
I restate - I agree with the intent. I endorse the intent.... the execution is crap.
Elliot Silver (Jan 20 2022 at 17:54):
I think Lloyd point is that this has been stated (and I’d heard it before, even as someone not following FMG or Vocab that closely), it’s just not been in the automated tools.
Lloyd McKenzie (Jan 20 2022 at 19:44):
I don't think we can kill the tracker items that have been raised. They're valid and the actions need to be taken. The fact the QA isn't yet hollering about it doesn't mean it isn't valid for a human being to holler about it.
John Moehrke (Jan 20 2022 at 23:51):
I am not asking to kill them... that was someone else... I am asking for us to have a coordinated and consistent handling of them. More specifically, instructions on how (process) to handle them. so that workgroups don't all need to spend energy inventing their own response. Meaning, your clarifications on this thread have helped me, and I will help Imaging workgroup that was one workgroup hit by this... other workgroups should benefit from this
Lloyd McKenzie (Jan 21 2022 at 01:12):
We can discuss plan and communication approach tomorrow Q4, time allowing
Eric Haas (Jan 24 2022 at 19:21):
was it discussed?
Lloyd McKenzie (Jan 25 2022 at 00:33):
Yes. I'm putting together a presentation that will go out at the next co-chairs meeting that will outline the expectations for all work groups, along with the expectation of when the FMG will start enforcing the rules (Sept. ballot cycle). If you have balloters driving you to fix sooner, you'll need to negotiate with them if you want to delay until that cycle - given that you'll have to do it eventually regardless.
Last updated: Apr 12 2022 at 19:14 UTC
