Meeting minutes
Changes to document sections
<maryjom> w3c/
Mitch: I have a couple of items
… First, getting structures from text was programmatic [...] are by definition accessibility supported
… I won't worry about the differences between text and programmatic if it is real
… Jonathan Avila said he thinks it is not true in this case, especially when it comes to AT scripting
… I am not sure what the reality is
Janina: Sometimes people will try to make some sort of GUI from text interfaces
… It is not that easy these days as you can control the grid
… You no longer have programmatic exposure for these structures, you have to guess
… As opposed to GUI where hopefully that is exposed to the X Tree
Jason: In Emacs ou can identify tables, so much is programmaticaly available there
… There are options for that
Janina: It is its own world on Linux
Mitch: Are you talking about a text-based Emacs or a graphic-based?
Jason: It does not matter. It is extending Emacs in its own language so that it states everything, generates audio cues, it is pretty much its own interface
… It looks at what happens inside Emacs (functions, variables, messages) and produce a very efficient, customized interface
Janina: You can build a machine without installing anything else and have full Emacs functionality
Mitch: The comment about the general statement not to worry about tabular data because there is technology that tqakes care of it, isn't it overboarding it?
Jason: ATs have to be able to deal with it and there is precedence that they can do it
Janina: I can try to capture this nuance
MaryJo: IF there are things that an author can do that can break things, we could talk about ways for these applications to support accessibility
Janina: What is the automagic within Emacs?
Jason: Can't think of any. They can offer commands to manipulate tables rows, for example
Janina: It's just pulling data from /sys afaik
Jason: I have not looked in detail lately
… It does not have JavaScript and DOM support though
… IF you can write code that identify where columns are and their info, I think you are under control
Mitch: If you have rows and columns of text, can you identify cell coordinates?
Janina: Cat can do that
Mitch: Is this tool taking plain text on a character grid for column and column header and presenting cell-related information?
Janina: Not if it's done with spacing probably
Mitch: I saw a new command line tool failing because they wrote text as rows and columns. And then they realized there is no need for them to do it that way, they can provide an option for text to be linearized
MaryJo: There are a few edits that people have suggested.
Janina: Please take them, and I will work in another PR that addresses these broader suggestions
MaryJo: I'll add edits, merge, let you know when that's done, and then you can continue to see if you can bring some of today's conversation in
Mat: You can just keep adding commits into a PR branch and it will include these changes
Janina: I am good either way
MaryJo: I don't mind us continuing to work on this one, this is still a draft. The TF is not going to look at it until it is ready
Meaningful Sequence
MaryJo: There were different answers for Meaningful Sequence for terminal emulators
… Some tools have complicated GUIs
… Is it a Yes or a No for terminal emulators?
Mat: Just looking through the minutes
Jason: What was the potential isssue?
<maryjom> Minutes from last week: https://
Mat: Columnwise output mainly
… Jason said he would probably tell it to use linear output, but does not necessarily have to be there in the tool
Jason: We actually agreed on that
Mat: I want to make sure we don't treat this different than table guessing, which is about heuristics mainly
… Detecting tables may be easier than detecting columns. Columns would be tedious sometimes
Mitch: I think we had decided a couple of meetings ago that the UI of the terminal emulators was not in scope for this sections because that is just regular software
MaryJo: These are a special kind of bridge where you do not necessarily have all the information on the UI
Mat: Happy to go with what Jason asserted for this
Jason: If we come up with a good example I could consider changing
Mitch: I could give one. Columnwise with annotations at a word level that are at some distance
Jason: That would go for the terminal applications. There are two entries in the sheet.
Mat: That is a good point. I've had it myself, but that is application
Janina: Sometimes terminal programs can automaticaly control the errors generated by these applications. It is not the same thing
Jason: Maybe we need to clarify that we are covering both
Janina: Agreed.
Sensory Characteristics
Janina: I can't remember seing that colorization
Jason: I think they use the colors but they are not consistent, thus they don't refer to them in the text. Depending on what terminal you use they might be different
Mat: You can make colored text in the terminal. Somebody could come along and say "Refer to the bright/red text". It is probably unlikely, but you can specify an alternative palette for the terminal
Jason: I think we are all on the same page
Orientation
Mat: We agreed on the terminal emulator. I was not 100% convinced on this one. From the terminal apps perspective, they only output to a screen with fixed coordinates
… This can overlap with reflow, although most modern programs are well aware that you could have narrower viewports
MaryJo: Orientation was for mobile. Do people use termianal emulators for mobile?
Janina: Yes
MaryJo: It's up to the application, not to the content, to adapt, right?
Mat: There are signals you can get as to whether the viewport has change
… If the app is running and gets told that the viewprt has changed they could react to that
Jason: And that's probably covered in reflow
MaryJo: Some apps don't even allow you to change orientation
… I think that is up to the app and the output would be probably unaware
Jason: The terminal emulator will be ocnscious of the change and whatever runs on it will be adapted to the new rows and columns
… You could say that the emulator needs to react to these signals and the output should be already ready for that because they need to comply with Reflow
MaryJo: I would say No, and then Reflow would be in charge of the viewport changes.
Janina: I would agree.
Mat: It does not make a difference to the outcome as we are covering it somewhere else
Jason: I am comfortable with it
Identify Input Purpose
Jason: I am pretty sure the SC refers to markup languages
JAnina: AT needs to pay attention to this
… Masking is an issue, you can hear the passwords as you type them
Mitch: I agree with what was said for WCAG2ICT in general
… Both mobile OSes are having these programmatic identifications for input. I am not aware of those in a text-based interface
Jason: For passwords and passphrases that is not echoed to the terminal, so AT could decide not to echo these either
… It does not have to identify the input purpose
Janina: That is the correct engineering solutions
MaryJo: This is not an applicable requirement
Mitch: It could be hypotheticallyapplicable for future technologies
Jason: I think that is reasonable, any qualification should be in the notes
Resize Text
Jason: If it is a good case then we should change our answer
Mat: I may have a bigger font that gives me fewer columns, but again, this is Reflow
… Maybe we need to say yes but it is covered in Reflow
… I put Yes for the terminal emulator because they should allow you to enlarge the text
Jason: I think that is reasonable. You may be using magnification or have an OS text size where the text interface don't interfere with that
MaryJo: We will start back from 1.4.11 Non-Text Contrast
… Through to 2.4.7 Focus Visible