From The Linux Foundation
Revision as of 01:36, 14 May 2008 by Oedipus (Talk | contribs)

(diff) ←Older revision | view current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search

Open A11y Working Group Conference Call (6 May 2008)


  • Janina Sajka (JS/chair)
    • Stew Benedict, Linux Foundation (SB)
    • Pete Brunet, IBM (PB)
    • Calvin Gainsford, Novell (CG)
    • Kris Van Hees (KVH)
    • Gregory J. Rosmaita (GJR/scribe)
    • Olaf Schmidt, KDE Accessibility (OS)
    • Andrew Upgrove, The Linux Foundation (AU)
    • Willie Walker, Sun Microsystems (WW)
      • regrets: Karen Copenhaver, Gunnar Schmi-DT

For Reference

Preliminary Items

Approval of Past Minutes


CG: posted to accessibility list on change in governance in AIA; steering committee appointed -- held 3 elections in recent months; first to have members approve current steering committee (result: no change); second, voted for additional seats (result: yes), third number of additional seats (results: 2); taking nominations to fill 2 new vacancies by 21 May 2008

JS: good that governance model at AIA is open to reflect needs of constituents and members; good to hear; got email yesterday from Rob Sinclair asking about Linux Foundation Board of Directors, and Board of Directors election policies]; responded with my experience and a pointer to the Linux Foundation's Bylaws, and to the Linux Foundation's Technical Advisory Board (TAB); Rob Sinclair looking to see if what AIA was doing is similar to LF; good to know we are doing email ballots -- plan on doing same in AIA

PB: assume nominations will come from within AIA? -- probably several organizations that didn't join AIA -- wonder about discussion from those who would like to join, but haven't been invited

CG: i don't know -- i'm just the messenger, and please don't shoot the messenger -- Rob Sinclair is the person to contact to ascertain if AIA wants additional input

Meeting Minutes

Note: These minutes have not yet been finalized.

Topic 1: Intellectual Property Rights Policies

Special Guests: Andy Upgrove and Karen Copenhaver

For Reference:

JS: glad OlafS here; Olaf one of leading advocates and watchdogs on policies that keep us open and inclusive;

JS: (offers floor; no takers)

JS: Andy, concern about assert clause -- minimum standard for whether promise would apply -- bit of tension between open source and open standards -- very important distinction; LF similar -- have Linux Standard Base (LSB), which has conformance requirements in order to be LSB compliant, certified, etc.; community promise based on that seems quite orthogonal; want to do if setting up mechanism if trying to get people to meet to set up standards of good practice; can take code from LSB and use either a little or a lot -- if publish, have to send back to LSB for confirmation (put in link) -- seems different than defining a standard

AU: for those not on call last week: core principle of GPL and similar is that can change anything you want in software -- makes sense, b/c part of value of collaborative work is to build on others' work; standards: restraint not to change - otherwise wouldn't be a standard and wouldn't foster interoperability; is a hammer-to-apple relation between open source software and open standards; gives rise to difference in IPR commitments; standards (pre-date open source community organizations): historical thinking was don't pick anyone's patent pocket, but provide enough benefits that patent owner has rights to share with others to promote interoperability; trying to balance what patent owner gives up; open standards: implement standard and no more (if don't implement entire standard, don't achieve interoperability); core document and extensions need conformance statements declaring what must be implemented in accordance with RFC2119; can infringe on patent if use to build interoperable product;

AU: open source: what if want to implement standard in open source project; way to get win-win is to (predicated buy into open standards) have a non-assert covenant or non-patent license that includes no restrictions, but should be using to promote interoperability

AU: Linux Foundation: LSB lives between kernel and other parts of linux and application; if wish to achieve interoperability, must conform to standard on which original application is based; standards developers and open source developers need to be in constant dialog; need to sync with kernel developers -- standard and kernel so far apart, don't want to damage either but rather achieve interoperability

AU: brings us to where we are today -- is MS OSP promise good enough -- any holes? go far enough? -- non-assertion covenants are an evolving discipline: IBM, Sun, Nokia, Oracle all have made extensive linux or open source covenants; MS doesn't pass muster of Freedom Law Center; if invite AIA/MS into fold, want to ensure that limit of promise is sufficient to not damage any projects or interoperability;

AU: as noted, OSP relates only to non-commercial side of development; couldn't do RedHat for example any good; flunks overall GPL test;

AU: hurdles: 1) open source freeware ok; 2) settling IPR for commercial use; 3) could this be negotiated or is this all just a diversion

JS: anyone?

OS: interoperability of accessibility; collaboration essential; anything that would make it legally unclear if can be implemented in open freeware, then a non-starter; questions related -- diversion or good-faith action; legal side is major test of whether can collaborate and sit at same side of table; if MS willing to change the OSP to deal with our concerns, then that itself is a good sign of good faith and of a willingness to listen, but need to know how long such a change would take; what changes precisely are needed; related to technical topics - technical discussion is very much linked to software licensing

JS: still unclear what we have to look at -- any software that comes out of OSP would be licensed in a way we are comfortable with

JS: PeteB's comment from last week very pertinent - Open A11y projects are gaining momentum, why stop now for hope of something that may not happen

PB: MS asked the IA2 team to extend IA2 to include features in UIA, i.e. two new methods to map back and forth between MSAA IAccessible/ChildID and an accessible object (UIA or IA2) -- there are 3 different a11y systems: 1) MSAA (uses 2 parameters: IAccessible/ChildID); 2) UIA uses a single accessible object; 3) Ditto with IA2; to get interoperability between UIA and MSAA, UIA created pair of methods to map back and forth from a single parameter identifying an accessible object to the pair of parameters used in MSAA - MS asked that same be added to IA2 to map back and forth between MSAA and IA2 -- what kind of legal restrictions are on IAccessible2?

AU: if standards infringe on MS patent, have no assurances

PB: IA2 doesn't infringe.

AU: no promise equals no protection; if work you do on IA2 free-floating between MSAA and UIA, anything from MS in terms of promise is benefit, because you haven't tainted the starting point; pragmatic observation:

[AU dropped]

PB: methods to translate back and forth can be defined separately from the existing set of IA2 interfaces to protect existing IA2 from MS legal claims - there is only one way to express this mapping, i.e. one method takes two input parameters (IAccessible and ChildID) and returns a single reference (to a UIA object in UIA or an IA2 object in IA2) - but these methods existed first in UIA - and although there is only one way to do this we don't have a statement from MS legal staff that there is no patenter copyright issue.

CG: asked MS via email for comments based on last conversation; MS said "commented in bug report that it is ok"

PB: technical manager's ok, not an attorney's ok

JS: no formal statement on behalf of corporation -- seems in conflict with promise -- meaning of standard not yet defined

PB: why i asked for Andy's opinion

[AU back]

AU: if what you create is independent of what you've already got in IA2, won't taint IA2 itself; likelihood of suing very low, since request from them -- historically MS hasn't set patent traps for people

PB: could make separate interface

AU: business point of view: if you create something for MS, and the community is not comfortable with OSP, then you've help create sourceforge converter that makes everything "interoperable" but not reality because open source vendors distrustful; want to ensure that lots of high quality, free interoperable tools out there to build confidence and trust in extra layer;

AU: get the patent policy YOU want; practical result should be that everyone feels equally comfortable

PB: need to work at overall level first, then strategic level

CG: can give that feedback to MS -- just having developers manager saying it is ok is not going to get past community promise issue -- needs to be more

OS: especially because different promises in past -- practice turned out to be wildly out of sync with what was in promise; community trust VERY low; need to base on actual promise

CG: agree

WW: higher level question -- what do we gain by cooperating and what do we lose by not cooperating

WW: if so difficult to have discussion here why bother

JS: need to confront reality that answer might be why bother

JS: carrot is replicate FF experience

PB: one option is a common API that could be mapped-to -- write to new API that shared across platforms;

AU: another possibility: what's best from our point of view: 1) go full speed ahead and not be slowed down; 2) build better tool kit - bridges between linux and windows because of predominance of the latter; happy to work with AIA, but we want you to do your share of the work; we will submit our standards, collaborate with you so can interface with our work; build bridge from their side to span the river, rather than building from either end and hoping meet in middle; would be way to take risk out of IA2

PB: CG implementing UIA bridge for linux - same could be done for windows

CG: don't speak for MS or AIA; Novell trying to improve our own accessibility on linux and has benefits for mono, but think that they are looking for Open A11y to define the issues -- what is the work they need to do to satisfy compatibility so can work together; think that is what we are trying to discover -- what does MS need to change?

CG: may not be all MS looking for, but is definitely part of it

JS: API only or software -- software needs license that we can live with

CG: our intention to port UIA to linux is to go with MIT X11 -- needed most open license

CG: legal department didn't pick up issues; we intend to implement the whole spec, but what if someone takes pieces of code -- diff issue, and needs to be addressed

PB: is it open now?

CG: yes

WW: maybe confusing separate issues -- on linux base trying to play well with native accessibility and writing bridge to ATK -- MS does work to meet existing structure; other problem i perceive is tension between UIA and IA2

CG: understand

WW: AIA wants to push UIA as solution, while IA2 trying to become the solution -- does that characterize the problem?

JS: does to me -- lack of clarity in license where a lot of that rests

CG: MS coming to meet us halfway -- IA2 and UIA out of domain

JS: IAccessible2 is Open A11y's specification

CG: what does that have to do with linux?

OS: started working on that when we were under Free Standards

PB: IA2 and UIA similar in what need to accomplish; no AT with UIA support; similar to UIA to ATK bridge Novell is creating, there could be a UIA to IA2 bridge or a IA2 to UIA bridge or third holy grail API set that is commonality between the 2 - ATs then needs to implement the common one; those are our 3 options; windows world trying to figure out direction in which wants to go

CG: what is probability of IA2 being supported? common API for all -- probability of all three choices outlined -- application providers

PB: based on what saw at CSUN is a lot of people want to implement IA2; and a lot of implementation already exists, but MS carries a lot of weight and need to ensure all on same page; need to participate with them to get legal points in place so AIA and LF can participate in UIA/IA2/holy grail API work, where developers of all kinds are working on a common standard

WW: MS developed APIs in past haven't implemented fully in own products -- is this one of those things that will happen with UIA

CG: not implemented UIA in IE or Office; Office accessibility from MS participate in UIA specification group -- owned by AIA now; no idea what intentions are

PB: implemented in Silverlight, WinForms, and .NET, right?

CG: windows presentation framework is first level that will be implemented

WW: need to know the lay of the land

JS: other piece is LF/Open A11y own IA2 -- donated to FSGA by IBM; on fast-track to ISO standard; felt by all that that drives more support and participation with IA2; a lot of support and attention to ISO standards; don't want to have to put off progress of IA2 -- maximize work on IA2 and get submitted to ISO

PB: MS probably has roadmaps and designs for UIA - may be disruptive, so maybe is a downside to that from leadership point of view

WW: substantial difference functionality wise?

CG: very different; a lot of windows stuff mixed into; IA2 a lot more like linux

PB: the Novel UIA to ATK bridge could be expanded to map to IA2 or ATK/AT-SPI

CG: mapping provider-side to ATK; specifically to get WinForms accessible; story may change as spec implemented;

PB: gaps?

CG: yes; haven't done full gap analysis

JS: reasonable stopping point; not sure what the next steps are, but will keep investigating

[AU leaves]

Topic 2: ATK on D-Bus Update (conditional)

For Reference:

Discussion Points:

  • What applications need to be rewritten that will not be covered by CodeThink, e.g.:
    • gnome-speech, py-atspi, and others
      • How do we move these in parallel with the port to D-Bus?
  • How to test on small devices?
  • How can we use this work to expand the community of accessibility developers in Gnome, KDE, Linux, etc.?

Topic 3: SIG Updates (conditional)

Topic 4: Topic 4: Website Update (conditional)

Open Questions:

New Business

Identify Topics for the 13 May 2008 Teleconference

JS: like PK to give us a debrief from TITAC (put in link)

JS: probably take a break from IPR conversation for next week or so

GJR: perhaps PB can do an intro to IA2 for group

PB: could do in 5 or 10 minutes; worked hard so could put abstraction layer on top and get to MSAA or AT-SPI, but UIA so very different