Meeting minutes
Explain the metadata workflow in 2.2 · Issue #397
<AvneeshSingh> w3c/
AvneeshSingh: Thanks Rick for creating the image.
mgarrish: I hadn't seen the image initially when I created this based on the github submission review.
… I wasn't sure how to use this flowchart, there wasn't a strong narrative, only a description was what the text and what was flowing on the chart itself not how the metadata would actually do in systems. Unsure of what it was trying to show.
… maybe it should talk about the metadata either embedded / outside with ONIX. The description just not explaining it well.
rickj: I understand where you are coming from. Someone coming into this would try to use this flowchat to understand how the metadata flow. We were trying to show how this all the parts that work together.
… we need a better description to describe that.
mgarrish: Now this makes more sense. A workflow to me how the metadata flows through not how all the pieces of the ecosystem work together.
… maybe taking out the "Flow" might help or how we can improve the description. how to understand the parts of the infrastructure rather than a "work flow"
gpellegrino: this assumes focusing on EPUB. we may need to update the labels to be more neutral.
rickj: If we have PDF in the crosswalk then we can add that in this image as well.
mgarrish: EPUB specific the metadata delivery should be generalized and then we can push down the idea of EPUB / PDF, maybe at the processing level / spec level. generalizing it some more would be good.
AvneeshSingh: Rick/Matt/Gregorio sounds like you are harmonized with each other and can come up with an updated description / narrative.
Brief update from the meeting for exploring inclusion of PDF accessibility metadata in the guide.
AvneeshSingh: Action item for the them to update the Description / narrative.
AvneeshSingh: Had meeting with Duff Johnson about PDF and including that in the crosswalk of a11y metadata
AvneeshSingh: PDF folks are open to our suggestions. Rick suggested expanding the crosswalk, we currently have EPUB & ONIX. maybe a new column for PDF metadata.
… once we have that we can add a new techniques document.
… Duff though this was a good idea and he can bring it to his association. PDF/UA has a number of these accessibility features.
… there can be a way to embed ONIX codes inside PDF.
rickj: ONIX coding PDF can attach XML files declaration, they will create a new one for oNIX fields. There are 3 categories of PDFs UA2 and UA1 and what are "tagged" as accessible and what assumptions we can make from them. "PDF Declorations"
George: image PDFs no accessibility (for completeness take them into account).
… going to their working group, may take them a while to fill in this crosswalk, My impression was that might take more than a couple months.
… agree making the principals more format agnostic. So we wouldn't put in a PDF techniques yet.
AvneeshSingh: starting point of the discussion.
… to make principals doc more neutral, to have 1 more entry in the ONIX metadata to have PDF/UA
… we will eventually have a PDF crosswalk.
The title of the guide is updated.
AvneeshSingh: we have updated the title of this document.
<AvneeshSingh> w3c/
<AvneeshSingh> New title: Accessibility Metadata Display Guide for Digital Publications 2.0
George: created an issue about the title "user experience guide" and we ended up about the "display of the a11y metadata" we discussed if it should say "digital" so we made that obvious.
… user experience guide was referenced and so all of them have been updated.
AvneeshSingh: issue was open for over a month.
mgarrish: title I am fine with. the URL still has UX guide, should we update the URL?
George: also the repo itself has UX Guide.
AvneeshSingh: we should do that on our next public draft and we could have a redirect couldn't we?
mgarrish: We did set something up with W3C, we can keep that one for 1.0 but a separate one for 2.0 I haven't thought it through yet, but I am sure we can get it done, discuss with Ivan.
Harmonizing strings in principles, ONIX techniques and EPUB metadata techniques.
AvneeshSingh: there is a lot of work on this topic. editors can provide an update, and Rick has an issue #381.
gpellegrino: all the localization system we set up for the techniques are based on ids, so every string or chunk of string is defined by an ID. for each ID there are 2 versions "compact" and "descriptive" for long version. in the techniques we only use the compact version, and we display the ID near the string.
… this work was done while working on the techniques then went back to the principals document to reconcile all the IDs.
… while doing this we discovered that some IDs were missed in one or more of the documents, so we are making sure that the same IDs are used in all 3 principals / EPUB / ONIX techniques. using automatic scrips and an excel file.
… when all 3 are aligned based on the IDs & same strings then we can create the final JSON for each ID for both the compact and descriptive options.
… most updated version is a PR, inside the localization folder.
George: I have not updated the strings in the spreadsheet, but I do put the ID and compact and description string in my PRs.
<AvneeshSingh> w3c/
rickj: #381 has two things, just some inconsistencies which looks to be fixed by the recent work.
… we are also inconsistent in the examples, we typically use the compact version but sometimes we use the descriptive. we should point out why we did this and call that out
gpellegrino: I agree. thanks Rick for opening this issue. 1. provide a note in the techniques to explain what version we are using the compact version. 2. as soon as we have the IDs align we will make sure there the strings will align and only the descriptive versions will be in the principals.
George: I have a PR from issue #381. I have some questions. the issue #408 is almost complete.
… non visual reading has 4 items and you couldn't get to it, issue #367, so a lot of items will get addressed.
… #381 asks about things not shown. example WCAG-AAA, we were thinking that we don't expect to see come very often get AAA, we we want to take up the space to show these examples which we don't expect to see in the wide. Could say the principals only show a subset of all possibilities.
… thanks Rick for your feedback.
rickj: when do we expect the final JSON localization file?
George: I have next four days to work on this.
AvneeshSingh: I think 1 month.
gpellegrino: I agree
AvneeshSingh: we also have some conferences coming up ect.
rickj: if we don't have this done by the end of the year then Vitalsource won't meet the EAA deadline.
AvneeshSingh: We should accelerate this.
… make this a priority as best we can after Frankfurt.
George: everything associated with the principals a lot of little things but the biggest focus that the strings are all correct, between sheet and the principals and techniques. THen continue to work on the broader statements / descriptions etc.
AvneeshSingh: as soon as we are satisfied we get a first version of the JSON then we can refine the strings if needed, completed within a month.
… need to figure out the editors call with Frankfurt etc.
rickj: happy to help
George: Gregorio can you explain the spreadsheet and generation of the JSON file
<rickj> if there is time today, I would like to at least discuss why I filed w3c/
gpellegrino: I am running a stripe that will create a structured format of all strings compact / descriptive, the excel is just to cross check or each review to make sure the same IDs are used and the strings are the same with visual inspection with sorting of the same IDs etc.
George: do you have a flag to show discrepancies?
gpellegrino: the excel file will be the dashboard to ensure everything is correct.
AvneeshSingh: can this be automated?
gpellegrino: I am using excel to check to make sure the strings match
AvneeshSingh: can Rick's team review this at all?
rickj: just let me know.
George: what about AAA? and there were about 40 IDs that were not in the principals
AvneeshSingh: down to 20 or so.
… we don't need 100% coverage
<mgarrish> CharlesL: there are small little words "anded" together and those are not in the principles because of how the algorithms generate them
Charles: there are connective words for algorithms so there won't be 100% coverage between principals and techniques.
gpellegrino: we just want to make sure the main strings are aligned. but there may be others in techniques only.
rickj: Any other business #384.
George: I will put in a statement that we won't have all possible values in principals but are in the techniques.
Any other business.
George: what about there is something in ONIX but not in EPUB. I filed an issue.
rickj: refines attribute on date of certification. how to pull out the refines attribute, only the date has a refines discussed, but the EPUB 1.1. has other refines, but not in the techniques but EPUB 3.3 implies refines can be used anywhere.
mgarrish: I would assume for metadata guide we want to make sure it alines for 1.1 everything can have a refines. but for certification it makes sense. just making sure the 1.1 guide aligns
… I don't think you will have accessibility metadata use the refine. only in the conformance strings.
Hadrien: most common use case is for contributors for example, Japanese there might be script variants. TTS could be a way that a narrator is synthetic, so this could come up again.
… general support for common elements would be good.
AvneeshSingh: also came in the Braille metadata. which will require refines.
… see you all in 2 weeks.