Minutes for Accessibility IRC Meeting of December 16th, 2010

• When: Thursday, December 16th, 15:00 UTC
• Where: irc.gnome.org, #a11y

Meeting Log

• 20101216_log.txt

Agenda

1. Documentation (shaunm)
   • Update about what docs go where
   • Personas needed
2. AT-SPI/Atk API changes proposed (fer)

3. GtkTreeview refactoring

4. Unity update (pendulum or themuso)
5. CSUN (eeejay)
6. GNOME 3.0

   • Status of the migration to PyGObject: GNOME Goal, 626218

   • Other updates
7. Others
   • Thoughts on "Do we need more hackfests?"

Attendees

Piñeiro, Fer, Joanie, Mike, Mario Sanchez, Shaun, Joseph, Eitan

Topics Discussed

Documentation

Discussion: New Desktop Help for Accessibility

• Shaun has been working on stubbing out feature-based pages in the desktop help for accessibility support.

• All accessibility features will be included in the main desktop help, but modules which need more detailed information (e.g. Orca) can also have their own application help

• Joseph asked if the new help should match the new preferences gui, adding that it appeared to match already.

• Shaun concurred, but mentioned that there were things which were missing, such as the Locate Pointer feature, and increasing the pointer size. In GNOME 2, these are part of the mouse preferences, but they're really for people with vision problems.

• William Jon McCann is the one working on the New Universal Access Preferences dialog for GNOME 3.

• Fer suggested colorblind support as another missing item.
  • Shaun asked what we had for that.
  • Fer mentioned the colorblind applet.
  • Joanie added that Orca has optional support libcolorblind, but that it was not a blessed external dependency to her knowledge, and that she had heard from Fer that the library was currently not in good shape.

Discussion: Input Shaun Needs from Us

• Current state of Caribou, and features of interest to people who use switches.
  • Shaun pointed out that GOK had switch support.
  • Fer pointed out that GOK would not work in GNOME 3.
  • Eitan reported that he has been cleaning up Caribou's code, that scanning support is already in place, and that he's currently migrating to pygi.
  • Eitan added that Caribou does have initial switch support, but that it could be improved.
• A better sense of what kinds of mobility problems people might have and what solutions and features were applicable.
• Personas
  • Joanie said that writing some for blindness and low vision were on her to-do list, but could not do more than just help out with personas related to other disabilities.

  • Eitan pointed out that AEGIS already had a set of personas, which at least Joanie, Joseph, and Shaun were unaware existed.

• What, if any, other areas need their own help documents?
  • Shaun suggested magnifier and caribou as examples, along with other alternative inputs like Dasher.
  • Joanie inquired about Accerciser. Shaun responded that Accerciser should be part of the Accessibility Developer's Guide.
  • Fer pointed out Dots.

• To what extent should he include information about "GNOME-friendly" AT that is not part of GNOME, such as Mousetrap, OpenGazer, etc.

  • Eitan thinks these should be pushed for module inclusion if they are ready.

Conclusions

• Shaun does not currently need personas from us as he can start with the AEGIS personas.

• Shaun will send a list of questions to the list.

AT-SPI/Atk API changes proposed

Discussion

• Fer reported that there were two things he would like to change, but that he wanted to discuss these with the team and also see what other changes team members might wish to have made. He also had questions about API breaks versus API additions.
• Piñeiro suggested that in-depth technical discussions are best suited for the list. Fer agreed.

API Breaks versus API Additions

• Fer asked if we could change the Atk/AT-SPI API as GNOME 3 (Gtk) is doing so. And even if we can change it, is it worth doing so?
• Piñeiro asked if Fer was talking about Atk2 or just improving Atk.
• Joanie responded that she thought it was worth it, and wouldn't mind seeing improvements in both places, due to text-changed events already being broken in Java and, she thought, Gecko.
• Piñeiro asked if these problems were in Atk or in the current implementation. He also added that at this point we would not be ready in time as Gtk3 is already "closing the doors" and Gtk4 planning has already begun.
• Joanie would still like to see the changes made with steps taken to ensure backwards compatibility because we have problems in toolkits other than Gtk.
• Mike asked when API freeze was. Fer responded 31 January for GNOME 3.
• Piñeiro pointed out the potential for breakage and the need for further discussion. Therefore, we can start this work, but plan for including it only in the new Atk2 or later as it would not be sensible to attempt to do this for GNOME 3.
• Mike and Joanie pointed out again the non-GNOME toolkits and applications in which there are problems. Fer added that if we do not fix it now, we will have broken applications for many years.
• Piñeiro again asked about the nature of the problems (i.e. Atk versus the implementation). Joanie responded that the problem was twofold: The current Atk API is not ideal in and of itself. In addition, improving ATk would make it possible to deal more effectively with poor implementation elsewhere (e.g. Gecko's native caret navigation).
• Mike stated that he doesn't see any reason not to add things, but that we need to preserve backwards compatibility. Piñeiro agreed.
• Fer proposed that the plan could be to add new things and mark the old API as deprecated in order to remove it in the future. He indicated that he would send mail to the list proposing addition of new signals/prototypes for text-inserted, text-updated, text-removed and selection-changed. In response to Piñeiro's inquiry, Fer added that new signals with new names is the only way to not break API/ABI

Atk Hackfest?

• Piñeiro reminded Joanie that we put the Atk rework on our Roadmap for April 2011.

• Joanie proposed having an Atk hackfest which would not be, in response to Piñeiro's question, a general accessibility hackfest with an Atk component.
• Mike asked if we should do this as part of CSUN or at Boston Summit.
• Fer stated he was in favor of such a hackfest, especially including Java developers, Mozilla developers, and other interested parties.
• Joanie tossed out the possibility of doing it in Spain as many of the people who need to be at this hackfest live in Spain.
• Piñeiro added that really it would be an Atk + AT-SPI hackfest, and that there is benefit in making the AT-SPI API more similar to the Atk one.

Conclusions

• We cannot break Atk API now due to the risk of breakage.
• It should not be a problem to add things and mark methods/signals as deprecated.
• This can be the initial work for Atk2 or Atk4 (the latter being Atk for Gtk4)
• We need to have further discussion about a specific Atk hackfest.

GtkTreeview refactoring

Discussion

• Piñeiro reported that GtkTreeview is being revamped for Gtk3, and that in theory things should still work, but that it had not been tested for accessibility.

• Joanie and Mike pointed out that GtkTreeview currently has performance problems which need to be addressed.

• Piñeiro agreed, but added that the thing to focus on now was to ensure the new version works.

Conclusions

• GailTreeview is not tested with the new GtkTreeview.

• Each of us with time should try to test it. (GtkTreeview has been integrated in the new Gtk.)

• Piñeiro will email Li Yuan about it.

CSUN

Discussion

• Eitan reported that:
  • CSUN is coming up. This year there will not be an associated hackfest; just a booth.
  • Because he has been so busy, Bryen has taken on a lot of the organizing. Thus he is not sure who else is volunteering to go, but he thinks it will be around four or five people.
  • He, Bryen, and Peter Korn will be presenting.
• Logistics/travel funding:
  • Eitan asked if he should approach the Travel Committee.
  • Piñeiro pointed out that when he and Joanie were working on the Team budget for the Foundation, they did include CSUN.
  • Joanie responded that while this was the case, they had not budgeted for four or five people; they had suggested funds for only three.
  • Piñeiro added that the budget was stuck as they were waiting for a final roadmap. Joanie pointed out they already have our roadmap.
  • Eitan mentioned that he thought there was someone from India who might be joining the team at CSUN, but he could not recall the name.
  • Joanie asked if this person expected funding from us; Eitan did not know.
  • Joanie responded that staffing a booth is not something that requires international travel, which tends to be costly, and that given our extremely tight budget we should fund active team members.
  • Eitan then mentioned the individual's name was Rohan and, in response to Piñeiro's question, that he was "Bryen's protege" and had done "a bunch of research things" but he did not know what specific GNOME project he might be working on.
  • (NOTE: After the meeting, Eitan announced that Rohan is studying at UCSD. This makes him local to the CSUN conference. Thus if interested, Rohan can be the third person and will not require travel funding.)
• Fer and Mike each mentioned that they might be going to CSUN.
• Joanie might as well, though she was going to speak on a GNU-initiated panel that no longer appears to be happening.

Conclusions

• CSUN is in March, we have space for two more people.
• If you want to volunteer, get in touch with Eitan.
• Eitan will provide more details next week.

Summary of Action Items

• Shaun will send a list of his documentation-related questions to the list.
• Fer will mail the list proposing the addition of new signals/prototypes for text-inserted, text-updated, text-removed and selection-changed.

• Team members should try to test the new GtkTreeview, which has been integrated in the new Gtk.

• Piñeiro will email Li Yuan about GtkTreeview

• Team members interested in going to CSUN to volunteer with the booth should get in touch with Eitan.
• Eitan will come to the next meeting prepared to give us more details about CSUN.

Next Meeting

• Thursday, 23 December, at 07:00 UTC in #a11y on irc.gnome.org