Meeting minutes
PLSI
Valerie: Curious about how to apply minimum role to PDF
Zakk: It's either minimal role or further guidance that needs to be given
Valerie: Minimum role is a concept in HTML-AAM, just not sure what this is used fore
… Minimum role is that if we put aria-pressed in something that is not a button, it'd become a button
… Same for draggable
… I could imagine on PDF-AAM, if something has an incorrect tag but it has an attribute then it might have a minimum role
JamesC: Minimum role [...] when certain conditions are met that are specified in the HTML-AAM specification
JamesC: Zakk, you may want to skip through and find all the instances of minimum role inside that document
Zakk: The intent of what we are trying to say we have conflated some things between PDF and HTML, not sure if this concept indeed is truly applicable to PDF. We don't have dynamic semantics with our structural elements
Valerie: This may be more suitable for the "repair" concept, where the browser would expose the correct role from guessing based on the attributes added by authors
JamesC: In ARIA we have a "generic" (shouldn't be named). If that's the case it may be more appropriate for the repair mechanism to be triggered
Duff: We don't have such a thing on the PDF. This seems more like a fallback
… Zakk: We don't have those generic containers. Each of ours have their own explicit semantics
… Group and section are the only structural elements
Duff: We talked about this before because we didn't know better. The only place we seem to be able to put them is on the generic role, because we wanted to have a conversation as to how these structural concepts are expressed
Zakk: This is an artifact of using HTML-AAM as our boilerplate. We interpreted minimum role differently. I think this section needs to be reworded. We could probably remove "minimum role" as it introduces confusion
… All our structure types are explicit and we will be providing the mappings either through guidance or through the existing Core and HTML mappings
… Custom structure elements would be a malformed PDF according to the specification
JamesC: Browsers do this all the time, figuring out what authors mean to do, and they tend to agree on how to repair author mistakes
Zakk: Once all of our current structure elements are mapped still the concept may be out of scope. Browsers can still make guesses based on the guidance
Zakk: In our own specs we avoid processor guidance, but this AAM is effectively processor guidance
… Is this a case where it's appropriate for us to get into error handling?
Zakk: Point taken, I think that's a v2
Duff: Agree for v1 we could avoid it by going through all stuff that maps clearly
Zakk: +1
Paul: There'll be much less error handling when we get to closer mappings
Zakk: Take away here is -- intent needs to exist but we need to reword it
Paul: in 3.5.1 we are starting to get into computed roles. I couldn't wrap my head around what folks were talking about. We could say for those that don't directly expose their semantics, see the tables below
Zakk: These are also describing the API values. This section was describing what those table roles do in our document
… We still have to work with these API authors for custom guidance for these documents
Duff: Should we select PDF-centric semantics and drive it through a process?
Zakk: I think we have some in there, I tried to include an example in one of our tables
Zakk: the proposed mapping for role document child we map to generic. That seems like one that we could pick, similarly to document fragment
Zakk: How could we reach out to platform API owners?
JamesC: There are issues open to platform maintainers, they get assigned to the relevant people
Duff: Maybe we should study that model, and then pick on a couple that are high priority, post them, and see what happens
Valerie: I think you should open an issue for each one individually to the PDF-AAM repo
JamesC: You could open the issue with several ones and then as we get closer to FPWD we could focus more on individual issues
Duff: Mission for us is to come up with that (known unknowns)