Hi Bill, Pete Brunet wrote, On 06/28/10 22:48: > Hi Bill. The NVDA guys are right, i.e. IA2 is defined as a COM interface > using COM things like IUnknown, HRESULTS, BSTRS, VARIANTs, COM's > definition of arrays. Code for Linux should be using ATK/AT-SPI.
COM is one thing, the other difficulty that I see is that IA2 is not a complete Accessibility API, but just the "missing part in MSAA/WIN32". So if you want IA2 on non-Windows, you also need to offer MSAA, and maybe even some other random WIN32 API. Not sure if MSAA even had a license that would somebody allow to do so... >>the IA2 documentation says pretty clearly that it is intended to also > run on Linux. > Where is that? I need to fix that. > > Open office implements UAAPI (UNO Accessibility API). I don't remember > if there is a bridge from UAAPI to ATK. There isn't one from UAAPI to > IA2, at least at this time. OOo has a native UAA/ATK bridge since OOo 2.0.1. And the UAA/IA2 is finally (active) work in progress now :) > I remember Harald did something when IA2 first came out to use IA2 (and > MSAA) on Linux. I don't know any of details. They might have > transposed the IA2 IDL into something suitable for Linux or perhaps to > an intermediate form that could then be bridged to the real IA2 and to > ATK/AT-SPI. It's been a while that I heard about this, maybe they stopped working on it. This was before Nokia bought Trolltech/Qt. They simply wanted to make porting of Qt Accessibility a little bit easier with this. I never thought it would be a good idea doing it this way... Malte. > > Pete > -- > *Pete Brunet* > > a11ysoft - Accessibility Architecture and Development > (512) 238-6967 (work), (512) 689-4155 (cell) > Skype: pete.brunet > IM: ptbrunet (AOL, Google), [email protected] (MSN) > http://www.a11ysoft.com/about/ > Ionosphere: WS4G > > Bill Cox wrote: >> I've heard conflicting descriptions of what IA2 is good for. Some >> NVDA guys seem to think it's a Windows only interface, designed to get >> around some limitations in the old Microsoft interface, and that >> because of COM objects and other windows-specific stuff in IA2, it >> will never run in Gnome on Linux. However, the IA2 documentation says >> pretty clearly that it is intended to also run on Linux. >> >> What's the actual case? Does it make any sense to port IA2 to Linux? >> What would the game-plan be, and is anyone actually working on it? >> Exactly which applications should we consider accessing through the >> IA2 interface? >> >> Thunderbird and Firefox both have better maintained and tested IA2 >> interfaces than atk interfaces. It might be a pretty good thing to >> access these applications through IA2. I hear mixed stories about >> OpenOffice, but a similar argument may apply. >> >> QT is a different story. I've heard they don't use IA2 in Windows, >> and that the IA2 interface is their effort to support Linux. Is this >> the case? The lack of an atk QT interface may be the single strongest >> argument for supporting IA2 in Gnome. However, if the QT IA2 >> interface is unfinished, and only meant for Linux support, wouldn't it >> be simpler to modify it to use atk, rather than write an IA2 to at-spi >> plugin? >> >> So, in short, exactly what is the vision of IA2 for Linux? >> >> Thanks, >> Bill >> _______________________________________________ >> Accessibility-ia2 mailing list >> [email protected] >> https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2 >> >> > > ------------------------------------------------------------------------ > > _______________________________________________ > Accessibility-ia2 mailing list > [email protected] > https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2 _______________________________________________ Accessibility-ia2 mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2
