Stream: fmg
Topic: IG hosting arrangements
Grahame Grieve (Jul 18 2016 at 22:17):
We need to talk about IG hosting arrangements.
Grahame Grieve (Jul 18 2016 at 22:17):
with regard to published IGs, there seems to be consensus that they'll be published at http://hl7.org/fhir/[code] where [code] is chosen by FMG for each IG that HL7 ballots
Grahame Grieve (Jul 18 2016 at 22:18):
this happens to be where they already live, but in the future, instead of being published with the ballot package, they'll be published to that directory directly, and each will have it's own version history like the standard
Grahame Grieve (Jul 18 2016 at 22:19):
I would like FMG to confirm this
Grahame Grieve (Jul 18 2016 at 22:20):
The second question is what to do about publishing the trunk version of them. I'm not sure what the arrangements should be, but it appears to me to be technically infeasible to host them at the same relative location in the current build
Grahame Grieve (Jul 18 2016 at 22:21):
I talked to @Josh Mandel about moving the current build to http://hl7.org/fhir/current, but that has unhappy security consequences compared to the current build
Lloyd McKenzie (Jul 19 2016 at 00:45):
I'd like to see the US-realm stuff either as "us-code" or "us/code", (I'm guessing the former will be easier for you)
Paul Knapp (Jul 19 2016 at 00:52):
So if there are mutiple realms with IGs on FOO there will be foo, us-foo, zz-foo etc? Wouldn't something like foo, foo/us, fool/zz better match the naming patterns?
Lloyd McKenzie (Jul 19 2016 at 00:57):
There's no coordination mechanism in HL7 that has ever allowed for comparable IGs developed in independent realms, so I don't see foo/us or foo/zz being terribly useful.
Lloyd McKenzie (Jul 19 2016 at 00:58):
Different decisions will almost certainly be made about workflow, communication paradigm, scope and a whole bunch of other things. And the granularities are likely to vary too. (One us guide might correspond to 2 CA guides or vice versa.)
Paul Knapp (Jul 19 2016 at 01:01):
Although I think it is reasonable for realms to have different IGs and for them to be published on the HL7 site it seems you don't?
So then why would you have US realm ones??
Lloyd McKenzie (Jul 19 2016 at 01:18):
We have to publish HL7 ones - there's no HL7 US organization to post under. However, my understanding is that we will also host IGs from other jurisdictions that ask, so it's not only a US-question
Grahame Grieve (Jul 19 2016 at 01:20):
we don't even know, right now, which IGs are realm based. Cool huh?
Grahame Grieve (Jul 19 2016 at 01:21):
anyway, this is a fun discussion, but a place for the current build of IGs is the pressing issue. and this week, it's pressing indeed
Lloyd McKenzie (Jul 19 2016 at 01:29):
Do you think prefixing non-Int'l IGs by realm is inappropriate? In theory, we should know for HL7.org stuff by whether the corresponding PSS is realm-specific.
Grahame Grieve (Jul 19 2016 at 01:35):
I don't know what's appropriate. It's just that I don't know whether the IGs we have are US specific or not
Lloyd McKenzie (Jul 19 2016 at 01:58):
Lynn should be able to tell us for everything that's balloted
Grahame Grieve (Jul 19 2016 at 02:13):
well, there's a difference between balloted and actual. here's my view:
- DAF: DAF-Core is US specific, but DAF is not - just as interesting outside USA.
- SDC: nothing about SDC is US specific except it's governance.
- QICore - claims to be interational, but perspective is wholly USA
Grahame Grieve (Jul 19 2016 at 02:13):
I think they are all officially US guides
Grahame Grieve (Jul 19 2016 at 04:58):
ok, well, no takers from anyone. from now on, I'm going to host them at http://hl7.org/fhir/current/[x]
Grahame Grieve (Jul 19 2016 at 04:58):
I'll change when/if we make some other decision
Lloyd McKenzie (Jul 19 2016 at 13:38):
Sorry, I went to bed. DAF, SDC are both US. It doesn't matter who they're of interest to - they're defined as US specs and are published by the ONC. (SDC has some US-specific vocab bindings in the data element stuff.)
Josh Mandel (Jul 22 2016 at 15:42):
As y'all may have seen at https://chat.fhir.org/#narrow/stream/implementers/topic/git.20for.20IGs : we have >6 volunteers who'd be willing to provide support with git/github. I'm not saying "the support problem is solved", but I'm reassured that we have a good base for community support.
Lloyd McKenzie (Jul 22 2016 at 17:04):
I presume we'll continue using SVN for FHIR core, at least for now?
Josh Mandel (Jul 22 2016 at 17:38):
Yeah, I didn't think that was up for discussion (though if/when it is, I'll be glad to discuss :-))
Grahame Grieve (Jul 22 2016 at 19:41):
one thing at a time.
Josh Mandel (Jul 28 2016 at 02:31):
Following up, I have some basic infrastructure working to build github-hosted IGs and publish the result, by automatically running Grahame's IG builder and stitching a few components together. Source for the project at https://github.com/jmandel/fhir-ig-builder
Examples:
| IG on GitHub | Published |
|---|---|
| https://github.com/test-igs/simple | http://ig.fhir.me/test-igs/simple |
| https://github.com/test-igs/daf | http://ig.fhir.me/test-igs/daf |
This infrastructure encforces the "org/repo" convention by forcing the published URL to map 1:1 with the GitHub URL.
Last updated: Apr 12 2022 at 19:14 UTC
