GSoC 2013 Updates

Formal progress updates and discussion occur each week during the GNOME Accessibility Team meetings.


19 September

Full log

Minutes


12 September

Full log

Minutes

  • A patch which demonstrates the working focus and caret tracking integrated into the magnifier.js work, has been accepted to GNOME Shell master after a freeze break request by Alejandro. Thanks Alejandro.
  • LINK: https://bugzilla.gnome.org/show_bug.cgi?id=647074

  • Magdalen has been updating the wiki to explain the work
  • https://wiki.gnome.org/SummerOfCode2013/Projects/MagdalenBernsCaretAndFocusTracking

  • Magdalen has been working on the release notes with Juanjo and is interested to see what he thinks of the last draft she emailed.
  • Magdalen has carried out some tests to check performance degradation caused by the atspi init but after seeking advice from Mike Gorse she feels the test is not complete enough for the results to be reliable: He explained There's a call to atk_bridge_adaptor_init in main.c and that she could time that with and without there being a call to atspi_init()
  • Magdalen will be working on the ui today as advised by Joseph. She also promised Jasper to update the magnifier and tidy it up so has to put some thought into that.
  • Magdalen goes back to uni next week and she wants to look into click lock sometimes soon.
  • LINK: https://bugzilla.gnome.org/show_bug.cgi?id=707666

  • Magdalen also updated https://bugzilla.gnome.org/show_bug.cgi?id=707693

  • There are some doubts about bug 707666. During the following week, we will try to answer them.

  • Joseph received an email from the GSoC admin outlining the end of the program, and how students are to submit their code.
  • Josephs wants to make sure Magdalen is aware of this.
  • http://www.google-melange.com/gsoc/document/show/gsoc_program/google/gsoc2013/codeguidelines

  • They suggest that "pencils down" is Sep 16.
  • Quoting from the email: "We suggest that students have completed their projects by this date and spend a week writing documentation and wrapping up their projects. We require that students stop all coding on 23 September. "
  • "this date" is Sep 16.

5 September

Note: There was no meeting this week because the team was busy putting out last-minute 3.10-release fires.


29 August

Full log

Minutes

  • End of last week, Joseph worked with Magdalen in IRC preparing the latest patch for the tracker + magnifier.
  • Mostly were working on the tracker code at that time. More specifically Joseph was asking about that.
  • Magdalen posted a patch to bugzilla on Monday.
  • Jasper reivewed and Joseph reviewed.
  • Magdalen has had a lot of problems testing the files and hopes that the problem Piñeiro found will make testing easier from here.
  • Magdalen intends to get the next patch out as soon as possible and for it to be fully working.
  • Magdalen is not sure where the current work sits in the timeline.
  • https://google-melange.appspot.com/gsoc/proposal/review/google/gsoc2013/magpie/1

  • Regarding when/how to attach patches:
    • Bugzilla is a tool to report bugs, and provide patches that solve the bug.
    • Usually, those patches are finished (so working) and tested
    • If that is not the case, but the developer things that it is worth to upload it, the usual is adding a disclaimer.
  • ACTION: Magdalen is going to apply Mike Gorse's patch and start testing the work and then make the last changes that need to be done to get the patch ready.
  • There are a couple of warnings in the magnifier that also need to be looked at.
  • Magdalen is not sure whether to fix them separately from the magnifier work.
  • Steps recommended by Joseph:
    1.  Get a app that supports focus navigation via the TAB key.
    2. Start the magnifier.
    3.  Use the TAB key presses to move focus around that app.
    4.  Does the focused object scroll into view?  If not, it isn't working.
    5.  Drop into a terminal and change the tracking gsettings  — switch between centered, push, proprotional.
    6.  Does the mode of focus tracking appropriately change?
    7. Find an app that exposes caret motion through the a11y API.
    8.  Repeat steps 3, 4, and 5. with the text caret insertion point.

22 August

Full log

Minutes

  • This week was very unproductive and the focusCaretTracker and magnification edits did not make 3.9.90
  • ACTION: Magdalen will make sure the changes are in time for 3.9.91
  • The delay was due to a number of factors:
    1. The information Jasper gave her about how to run gnome-shell in jhbuild was not correct and this messed up the symlinks to libaries that are needed. Magdalen is not sure why that happened but she suspects it was because Jasper having too many conversations at once.
    2. The discussion about whether to make the tracker a singleton or not went on for too long.
      Magdalen feels she should be involved in decisions about work that she has to implement especially when the decisions make things more difficult to implement.

    3. Magdalen broke her computer before guadec and when she finally got it working that was a terrible connection.
  • Magdalen apologises for the delay and has nothing further to add
  • LINK: https://wiki.gnome.org/ThreePointNine

  • LINK: https://wiki.gnome.org/ReleasePlanning

  • LINK: https://wiki.gnome.org/ThreePointNine/Features

  • ACTION: Piñeiro will confirm if in order to have Magdalen's work in 3.9.91, a feature freeze request is needed.
  • ACTION: If needed, as soon as Magdalen's work is ready, Piñeiro will make that request.

15 August

Full log

Minutes

  • Magdalen apologises about missing the last meeting as she was in GUADEC finishing off the gsettings patch.
  • Magdalen wants to thank the mentoring team for their help with the lightening talk at GUADEC. She felt she could have done it better on the day but it could certainly have been worse, and all things considered she feels that she let many people know that this project was important one way or another during the course of the conference. :-)

  • Magdalen spoke to Karen about replacing the out-of-date GNOME magnifier how-to on the BBC website after it is finished and Karen seemed to be very interested in helping get the change publicised.

  • In GUADEC it was possible to speak face to face with the developers of GNOME Shell and it was possible to gain an
    • understanding of what would be required for the patch. The following pieces of information were discovered:
    • The gjs exception experienced by Magdalen shown here was actually normal and was not related to the crashed experienced in GNOME Shell. Further investigation suggests that the GNOME Shell extentions developers work around the issue by adding the line: "module_autogenargs['gnome-shell']=' --enable-jhbuild-wrapper-script'" and then running gnome.

    • The following command must be used from now on: 'cd /home/dev/gnome/source/gnome-shell/; ./src/gnome-shell --replace'
  • Magdalen has put separate .jhbuildrc configuration files in github: One for each environment she has avalaible because they each seem to require slightly different tweaks in order to work well (e.g. WebKit can not be compiled in a reasonable time frame.

  • Magdalen has run the !focusCaretTracker changes successfully but has not finished all the tests due to (yet another) mac crash slowing things down. However gnome-shell is built now successfully on both computers!!!
  • The only problem at the moment is that the mac thinks it is monday janurary 1, 2001 and although it is possible to 'ping 8.8.8.8' the connection seems to be limited in some way on wired and wireless.
  • HELP: Does anyone know how to set the date via bash because Magdalen cannot rebuild gsettings and include the change while the git.mk is so many years in the future, it seems.
  • It is not urgent as Magdalen updated gsettings from desktop since they were pushed to master :-)

  • HELP: Magdalen asks if anyone can clarify whether monday is a hard or soft deadline?
  • LINK: https://wiki.gnome.org/ThreePointNine

  • Magdalen has Juanjo, Joanie, Piñeiro, and Ryan to thank for helping to get the commit pushed before the freeze.
  • Magdalen is aiming for magnifier and tracker commit to land on or before 19th August but is not sure if there's a contingency or not in case there are any problems over the weekend.
  • Magdalen has a lot to discuss with mentors after this commit gets out the way and some ideas for after the primary aims of the project are complete but that will have to wait till after the commit and this is just an fyi. :-)


25 July

Full log

Minutes

  • Joanie just created a page on our wiki which, through wiki includes, sucks in our GSoC updates from our meetings.
  • LINK: https://wiki.gnome.org/Accessibility/Minutes/2013GSoC

  • Joanie will continue to update this page as part of meeting minute prep/posting
  • Magdalen has updated the wiki but the files are on my computer which seems to have a compatibility problem associated with efi booting
  • Magdalen is working on the lightening speech and is not sure how to proceed due to time restrictions.
  • Magdalen is pleased to say that fedora is now running smoothly on her mac though so she is with computer.
  • Magdalen has been using gjs and gjslint to tidy up the files and and debug because they give more debug messages than jhbuild run gnome-shell -r, but she had an odd debug message with the magnifier even when trying on the gnome-shell version with none of her local changes.
  • Joseph and Piñeiro continued their discussion about the 5 issues re focus/caret tracking on the gnome-a11y-devel list.
  • Joseph has answers to all of the issues, and has a way to proceed.
  • Joseph needs to consult with Magdalen to somehow communicate these solutions within the latest FocusCaretTracker code.

  • ACTION: Joseph will find some way to put comments in the latest FocusCaretTracker code that speaks to the solutions to the five issues.

  • ACTION: Magdalen will talk with Joanie and Joseph after the meeting.
  • ACTION: Magdalen's village of mentors will help her structure her presentation content for the GUADEC ligtning talk.

18 July

Full log

Minutes

  • Joseph had an action to post the five issues regarding the focus/caret tracker to the accessibility development list. He did so: discussion thread

  • Joseph is rapdily converging on these conclusions:
    1. The tracker should not be a singleton, unless it's too costly to have multiple D-Bus registrations.
    2. It's still a question of where Atspi.init() should be called.
    3. Atspi event registration should happen in the constructor in the non-singleton case (see 1).
    4. The try/catch statements can be safely removed.
    5. There is no use case for distinguishing between 'object:state-changed:focus' and 'object:state-changed:select'.
  • Joseph has tried to keep up with Magdalen's work, but has found little time to do it this week.
  • But, he has at least responded to her emails.
  • Magdalen has updated the tracker.
  • Magdalen is working on the magnifier.
  • Magdalen has been updating Joseph via email about changes and asking questions about it that way too.
  • Magdalen missed an email about the meeting for OPW but should go to the next one (no date yet).
  • Magdalen will be doing a lightening talk on the project for GUADEC.
  • Magdalen needs to hand in the slides by 29th July.
  • ACTION: Magdalen is going to start working on the lightening talk.
  • Regarding GSoC evaluations, the mentors are waiting for questions/form to be sent to the group.
  • LINK: http://www.google-melange.com/gsoc/events/google/gsoc2013

  • July 29 mentors and Magdalen can begin submitting mid-term evaluations.
  • The deadline to hand in the evaluations is 2nd Aug.

11 July

Full log

Minutes

  • Magdalen put together a patch for the FocusCaretTracker, and uploaded it to the bugzilla. (Congratulations).

  • Jasper reviewed it, and made some interesting points.
  • Joseph responded to his review, and has been ruminating over it since.
  • Joseph believes the way to address them is to answer a few technical questions:
    1. Should the tracker be a singleton?
    2. Where should Atspi.init() be done? (probably in main.c).
    3. Should event registration with Atpsi happen in the constructor?
    4. Should clients have the opporunity to respond to focus event separate from slection events?
  • While these issues are discussed with the "experts", Joseph feels that Magdalen should proceed with changing the magnifier to use the tracker as is.
  • Magdalen sent a copy of APIs advice chat log to the mentors
  • At the moment it's in the constructor since update
  • LINK: https://github.com/thisMagpie/GSoC/blob/master/javascript/focusCaretTracker.js

  • Joseph feels that Magdalen should leave the tracker alone for the time being.
  • Magdalen sent (as she info'ed) an email showing an IRC log of the converstaion with API about where Atspi should be initialized (question 1. above).
  • Joseph responded to that this morning.
  • Joseph is throwing the question to the Atpsi experts: where is the best place to do this?
  • Regarding GSoC "adminstration": Joseph is going on a two week vacation at the beginning of August. If there is any GSoC stuff he must do, that should be discussed soon. Also, there should be a plan for some of the other mentors to take over in his absence.
  • Magdalen wrote another blog post but it was not very detailed so thinks she should talk more about the gjs work in shell soon.
  • Magdalen has got further on the magnifier.
  • Magdalen also uploaded the work to github with the focuscarettracker
  • Magdalen is not sure how to change from global.get_pointer() without changing a lot of the mouse codes in magnifier.
  • LINK: https://github.com/thisMagpie/GSoC/blob/master/javascript/focusCaretTracker.js

  • LINK: https://bugzilla.gnome.org/show_bug.cgi?id=647074#c18

  • ACTION: Mike will test AT-SPI's behavior when registering multiple event listeners in the same process and comment on the bug
  • ACTION: Joseph will summarize the 4 main questions about the FocusCaretTracker, and send that to the accessibility dev list.


04 July

Full log

Minutes

  • Magdalen stopped work with the tracker last friday and waits for mentors confirm whether or not a patch can be made from it.
  • Magdalen has been using g-ir-doc-tools to make documentation for atspi in javascript.
  • Magdalen is hosting the files for times when the documents need to be discussed with mentors.
  • LINK: http://gnome.thismagpie.com/atspi.

  • Magdalen filed bug 703422 about documentation for atspi in javascript which Piñeiro reviewed and spoke to her about.

  • Magdalen started working on magnifier.js

  • Magdalen asked Joseph to break magnifier up so that each name space has its own .js file because she is getting confused by the amount of stuff in there but Joseph said doing that would be too much of a distraction.
  • Magdalen already edited gsettings before speaking to Joseph about it.
  • Magdalen made a video for mentors to check focus/select tracking is doing because she could not be sure what they meant about the log from focus/select.
  • Joseph worked with Magdalen at the end of last week to check the tracker code and its logs to make sure it was doing the right thing.
  • Joseph still needs to look at the final code on Magdalen's github, but he believes they arrived at the point of making a patch.
  • Joseph sent out email about next steps vis-a-vis integrating the tracker with the magnifier.
  • Joseph's advice is to leave the gsettings aspect aside for the moment, but notest that it will need to be address as part of this GSoC.
  • LINK: https://github.com/thisMagpie/GSoC/blob/master/javascript/focusCaretTracker.js

  • Magdalen also added jhbuild configurations to github.

  • Regarding spliiting up magnifier.js into separate files, not only is it a distraction, but it is wrong. It is wrong because it fragments the "Magnifier" namespace.
  • Joanie is wondering if anyone knows of the top of their head when the evaluations are due?
  • ACTION: Joanie will look for the date and procedure for mid-point GSoC evaluations.
  • According to the GSoC schedule, mid-term evaluations can be submitted starting 29 July with a deadline of 2 August.

  • Joanie will be traveling on the 29th so she'll either submit it later, or she'll ask another team member to do so
  • ACTION: Joanie will lead the coordination and filling out of evaluations.

27 June

Full log

Minutes

  • After the Magdalen's success last week, Joseph noticed the tracker was too public.
  • That was useful for debugging.
  • He suggested a way to hide most of the Atspi registry code from the outside world.
  • Clients just deal with standard signal handling now: connect(), disconnect().
  • Joseph will read the logs and see if all is okay.
  • Magdalen did send email about not receiving focus events for tabs.
  • Joseph forwarded that email to Joanie, Piñeiro, and Juanjo for their input.
  • "We" need to investigate that issue and see what is going on.

20 June

Full log

Minutes

  • ACTION: Joseph to review Magdalen's latest focus caret tracker code and possible suggest changes.

13 June

Full log

Minutes

  • LINK: https://bugzilla.gnome.org/show_bug.cgi?id=647074#c11

  • LINK: https://bugzilla.gnome.org/show_bug.cgi?id=672325

  • Joseph and Magdalen are going to keep freeze work on bug 681276

  • Magdalen will keep the next focus and caret tracking patch work for bug 647074.

  • There is an exception thrown when get_role_name is called and it is a 'timeout at dbind'.
  • Magdalen and Joseph are going to speak to mgorse about the dbus log.
  • Magdalen added the blog.
  • Another error occurs when calling get_component_iface(), which appears to be different from the one illicited by get_role_name.
  • get_component_iface() is more important here with respect to the magnifier. If that fails, you cannot move the maginified view to that component.
  • The dbus log contains a lot of stuff.

06 June

Full log

Minutes

  • Magdalen presented an update of her work.
  • She provided a feed if we want to follow her work: http://feeds.feedburner.com/thisMagdalen/GNOME.

  • She summarized her status as "right now I think I need to go through the js file with clown and identify the next plan of attack."
  • Joseph has read some of the GSOC emails from Magdalen and others.
  • Joseph has responded to Magdalen answering questions about magFocusCaretTracker.py, and supplying her with a clean copy of the focusCaretTracker.js
  • Magdalen updated code as well: https://github.com/thisMagdalen/GSoC/tree/master/javascript

  • She also spent some time working through how to include the JavaScript with a jhbuild of gnome-shell.

  • Next step is to find time to meet with Magdalen and go over focusCaretTracker.js, and explain what it tried to do.

Accessibility/Minutes/2013GSoC (last edited 2013-09-24 13:00:37 by JoanmarieDiggs)