[MINUTES] W3C CCG Traceability Call - 2023-05-09

Thanks to Our Robot Overlords and ben_(transmute) for scribing this week!

The transcript for the call is now available here:


Full text of the discussion follows for W3C archival purposes.
Audio of the meeting is available at the following location:


Verifiable Traceability Task Force Transcript for 2023-05-09

  Orie Steele, Mike Prorock, Mahmoud Alkhraishi
  Our Robot Overlords and ben_(transmute)
  Mahmoud Alkhraishi, Russell Hofvendahl, Paul Dietrich GS1, Nis 
  Jespersen , Ben (Transmute), Orie Steele, TallTed // Ted 
  Thibodeau (he/him) (OpenLinkSw.com)

Orie Steele: 
Our Robot Overlords are scribing.
Orie Steele: 
ben_(transmute) is scribing.
<orie> rate limit exceeded
Mahmoud Alkhraishi:  All right welcome to Chris Cole May 9th 23 
  IPR note please make sure anyone can participate any substantive 
  contributions need to be people have signed the ccg work.
<orie> Seems it can be fixed by removing the github release
Mahmoud Alkhraishi:  Welcome to thr trace call, IPR note, in 
  order to make any contributions make sure you have signed the 
<orie> or by being smarter about how we do releasres
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  There are no pr's on trace-interop, there 
  are PR's on trace-vocab
Mahmoud Alkhraishi: 
Orie Steele: https://github.com/w3c/vc-jwt/pull/77
Orie Steele:  I spent some time on this, it's still upstream. So 
  i'm not spending time on that PR until i have a better idea of 
  the path forward. Mike P make a PR on VC-JWT that's blocking. If 
  that can go through it will help it along.
Mahmoud Alkhraishi: 
Nis Jespersen :  This is minor tweaks to the MTR after using it 
  with real data. A couple chemicals missing. The product 
  description is always a single product and a fix to the 
  description of measurements. And steel grade is a string.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  It looks like it has approvals, merging.
Mahmoud Alkhraishi: 
Nis Jespersen :  This is adding syntax sugar to credentialSchema. 
  Looks like it has tabs and spaces issues. I can fix and merge out 
  of band.
Mahmoud Alkhraishi:  This is a PR to make a credential which adds 
  an identifier from an external system. Think of a blackbody 
  system where the identifiers are internal.
Russell Hofvendahl:  I did something similar, but I have problems 
  with the complications, so I think I dropped this.
Orie Steele:  I think there are some requirements in what you 
  said that are really important. So I think that the id's you're 
  using are not a graph-node identifiers. I think you also want to 
  give a name to this identifiers. So I non-graph-node identifier. 
  You need an identifier can be any string and a name for anything. 
  And these are provided with schema.org
Orie Steele:  So why are we definition our own terms, when we can 
  use Thing?
Mahmoud Alkhraishi:  I was trying to use Identifier, I didn't 
  realize there was Thing. I can change this to Thing, and ping you 
  guys to review.
Orie Steele:  Also I think this is a "Thing" credential and just 
  make it generic. I could be useful.
Mahmoud Alkhraishi:  There are no PR's on trace-interop, this 
  week is trace-vocab week. Do we want to goto issues, or is there 
  anything we want to bring up?
Nis Jespersen : 
Mahmoud Alkhraishi: 
Orie Steele:  Can we address the scope issue on trace-interop. I 
  think GS1 is still blocked on that, and I wanted to check where 
  we on that issue. I think Chris came away with some action item, 
  but I don't remember the specific details.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  This is the roadmap, which I wanted to bring 
  up. What is out concrete plan to getting from where we are today 
  to a v1.0 spec to say it's not fully locked down, but it's 
  reasonably stable.
Mahmoud Alkhraishi:  1. Do we want to wait for a vc 2.0 context, 
  2. do we want to publish on the w3c? 3. how complete are we 
  really, how many more changes do we have?
Mahmoud Alkhraishi:  That way we can try to put down a deadline 
  around locking down v1.
Orie Steele:  This is aligned with the CD process. We should trun 
  off the github releases, since we're not using them.
Orie Steele:  As far as publishing them, I expect a high 
  availability server that hosts the context file and never changes 
  ever. The vocabulary side of it, it can drift around. I think all 
  we need to do is host our context file that's permanent and 
  stable. What you don't want is context drift. And we want to make 
  sure that's mature and not going to change/
Orie Steele:  The reality is that it's an on-going process and 
  it's going to change.
Mahmoud Alkhraishi:  I agree to all of that, let's talk about 
  getting to that small v1 that's stable that will not change. We 
  separate by verticals, and we say that review each one of the 
  credentials in that veridical , we look at the terms and say 
  these are signed off. And then when we look at those and make a 
  cut off date to say that we expect the context to be published by 
  a certain date.
Orie Steele:  To be clear, we publish a single context as v1, and 
  we build that from credentials we believe has maternity. And 
  everything else moves somewhere else.
Mahmoud Alkhraishi:  So we have a development version with 
  on-going changes. And then we have a fixed version with the v1 
  terms. Do we think this is a path forward towards a fixed v1?
Orie Steele:  We can commit to that on the Transmute side for the 
  steel-tech demo to make those permanent. We could do what GS1 did 
  where they split their contexts files and intentionally made them 
  very small for that one credential. For Paul, do you think we 
  should try to make them smaller or make one larger context?
Paul: One second?
Nis Jespersen :  As someone who is consuming the context, I find 
  it hard to use, as I would like to have everything in one larger 
  to context. I queued to say, we don't want to change, but can we 
Orie Steele:  That's a no-no
Paul: We broke it up with the way we have things today. In this 
  case it might make sense to have instances for steel or gas, but 
  then where do you put the common stuff. So that might make things 
  harder if you have a lot of common vocabulary?
Orie Steele:  This has come up a lot in the v2 credentials 
  context? Will the v2 define everything, or do we want to make a 
  more layered approach where it could be more files.
Mahmoud Alkhraishi:  Is there a way to set up n umbrella context 
  that refers to other ones?
Orie Steele:  You can, but it's not an elegant way to do because 
  the document loader is always chaining.
Mahmoud Alkhraishi:  We could do both, we could have industry 
  versus common, or one with everything.
Orie Steele:  It just speaks to when you build the context what 
  are you trying to do?
Mahmoud Alkhraishi:  I think from where we are today and the best 
  cut of V1, let's pick the credentials that are ready, let's put 
  them in a context with the terms that they need. If we find there 
  are issues and we need to have more layers, we can push that as 
  another version.
Mahmoud Alkhraishi:  Other than the context questions, is there 
  anything else we see as a deal breaker before we can pushing the 
Nis Jespersen :  There's the JSON Schema side as well, that also 
  needs versioning
Mahmoud Alkhraishi:  For that I guess we set up a folder for as 
  of v1 this is the state of these credentials.
Orie Steele:  I would publish all of the schemas both as a a 
  directory and a bundled archive. And we publish potentially 
  outside of github.
Orie Steele:  Or we can create a Github repository that is 
  immediately archived. But if people are getting them, Github 
  might rate limit, so we might want to have something that's 
  stable and reliable.
<russell_hofvendahl_(mesur.io)> Going to have to drop
Mahmoud Alkhraishi:  What i hear from Mike is they're ready to 
  go. On our end, we're ready to say these credentials are ready. 
  It looks like Transmute is ready to do that to. And we need to 
  publish static JSON schema files, and publish the context on a 
  high availability web server.
Orie Steele:  Thinking through the mechanics of this, we might 
  want to handcraft a version 1 repository of this in order to 
  confirm what it will look like to have a sanity check that it's 
Orie Steele:  We are thinking that everything is built, it's 
  going to be a superset of v1. So we make it a fixed version, and 
  we make it so you can't push changes to the schemas or context.
Mahmoud Alkhraishi:  At the start, we need to do a dry run of it 
  to make sure that nothing is wrong when doing that.
Orie Steele:  We're constantly doing that. We can say that what 
  we're committing to now is what we're publishing.
Mahmoud Alkhraishi:  For that window, i still think that we want 
  to look through every schema, every credential, every term. Make 
  sure required and optional fields are correct.
Paul: Is this going to be done in a branch?
Orie Steele:  We're going to do it on the current branch, we 
  going to have a cut off date for no more changes after a certain 
  date. We probably have to look at schema.org. Or the JSON schema 
<orie> sry i need to drop
<orie> good luck
Nis Jespersen :  How do we want to version this? The way many LD 
  libraries are versioned in with a year stamp, would that work 
Mahmoud Alkhraishi:  I think a lot of people hate the year. 
  Because it makes a lot of sense to normal versions. Because we 
  have 2018 v1 and it's everywhere, and then they're just switching 
  to V2. We can add that as a question on the issue.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  I think we closed 297 because of no 
  volunteers. And we left this open, but no one is raising their 
  hand for it.
Nis Jespersen :  We've gone another way on the observations. 
  We've sort of denormalized the observations, which is easier for 
  the common folks to use.
Mahmoud Alkhraishi:  Personally I've hated working with it as 
  it's just generic. And I find that alot of documents don't map to 
  a generic way of working with these. And if you have to do custom 
  logic anyways. So we might just close this issue.
Mahmoud Alkhraishi: 
Nis Jespersen :  For this, do we want to have an enum, or a 
  format to try and get people to use a country code. When I use 
  this, I find myself putting in the country name more often than 
Mahmoud Alkhraishi:  The scope of the issue was just to open an 
  issue on schema.org. Ben have you been typing in the country 
Nis Jespersen :  Yeah. it's just a string value and I find myself 
  typing in the name more often than not.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  Okay I think we're using country codes, but 
  it it's not working as intended we might pass on this.
Mahmoud Alkhraishi:  For issue 555 it's marked as ready for PR. 
  Do we have any volunteers?
Nis Jespersen :  For this, i think that trace-able presentations 
  has caught on. Is there any appetite for this?
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  So basically it has legs, Nis opened it. Nis 
  is it okay to close?
Nis Jespersen :  Yeah, close it.
Mahmoud Alkhraishi:  This issue to was to break it out into 
  smaller issues. If there are no hard requirements for a scenario 
  that we're using, i think we want to close it.
Mahmoud Alkhraishi: 
Nis Jespersen :  Agreed, I'll write a comment.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  I put a link to WebVOWL. There is an example 
  in a comment i wrote. I think it needs a server to run. You can 
  click one, see what it's touching, and drag it around. It's 
  pretty, but I don't know what it does.
Nis Jespersen :  I think the question originally came from Vlad. 
  He wanted to see where our terms defined. Which ones are we 
  defining ourselves, and which ones are we reusing? It's a sales 
  argument for the work we do here. I'm getting pushback from 
  getting called a vocabulary since people think we're trying to 
  re-define all of the terms. So this is good to say, "these are 
  the existing terms that we're already using"
Mahmoud Alkhraishi:  So what we need to say is that we're not 
  defining 600 terms, we're actually using 400 existing terms. I'm 
  personally not running into these problems, so I don't know how 
  useful they would be.
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  For this one, the readme's are way out of 
  date for the PR's and we could have one referencing the other. 
  And it might be easier to have one link to the other where  only 
  one of them is the source of truth. I notice that everytime i do 
  the hosting the IPR note is on the trace side on the not the 
  vocab side.
Mahmoud Alkhraishi:  This is a big picture issue about how do i 
  reference a different credential, or a product that is not in 
  this credential.
Nis Jespersen :  I think i suggested using related link.
Mahmoud Alkhraishi:  If i have an event credential and I need to 
  reference a product id. So I could put in a product link, but 
  someone changes that. So it has the same problem.
Nis Jespersen :  In this case there is an id which is did:web and 
  the related link is for the identifier for where do you need to 
  go to find out about it.
Mahmoud Alkhraishi:  So I would have a purchase order, you would 
  have related link. Wait, the example doesn't make sense. I want 
  to have something that's a hash link or similar.
Nis Jespersen :  Can we go through that with an example?
Mahmoud Alkhraishi:  At the very top, we have a uri that points 
  to a VC. And that works except that doesn't prevent anyone 
  hosting the data from changing the information that. So I have a 
  product, and someone hosting that changes it. My event will point 
  to the same thing, but the thing i'm talking about has changed. 
  So we need a hash link so things break when things have changed. 
  it adds to complexity. and it not standardized.
Nis Jespersen :  How does that work?
Mahmoud Alkhraishi: 
Mahmoud Alkhraishi:  You add the link, but you add a hash and 
  then end of the url, and when you resolve that url if the hash 
  doesn't match, you don't trust that resource. In the vocabulary 
  we call it an external resource. We're doing it for images, but 
  we're not using it for other things. But if it works for images, 
  it should work for anything.

Received on Tuesday, 9 May 2023 18:42:34 UTC