Meeting Minutes: November 8, 2006
- Janina Sajka
- Bill Haneman
- Gunnar Schmidt
- Olaf Schmidt
- John Goldthwaite
- Gary Cramblett
- Peter Ferrenti
- Larry Weiss
Bill- sounds like the simpler solution for FSG would be to do it as optional module.
Janina- I’ve learned that the way they usually do things is to be a candidate for one rev, then a optional module for the next rev. and then full status in the following rev.
Olaf- all the work to do the standardize has been done, did not want to discourage use. LSB does not standardize for anything that is not in all major distribution. All accessible by AT/SPI, that means that KDE3 can not comply with that. The platform is not necessary accessible with AT/SPI but if you
Bill- in what way is that a down grade? That sounds like current status.
Bill- we have never made any requirement on application conformance. I am not aware of dependencies that AT/SPI has. Only application that speak AT/SPI or load the bridge will be visible.
Olaf- if we have it as an additional module, it could make it tougher on conformance. If we make it mandatory and we have not yet standardize, would that be
Bill- realistically , we can’t have a D-bus version for some time. Even if we had significant resources to put on it, it would take well over a year.
Janina- is anyone working on it.?
Olaf - even if it takes three years, we would have a standard. You could test against it. Distributions would like to ship the then available D-Bus version. If they want to re-certify, they have to ship the CORBA based version too. If KDE is part but optional. It may take a while, at some point D-Bus , all distributions that
Janina- is it really on the table to put it into 3.2? It sounded like they were skeptical if we are eventually going to D-Bus
Bill- what is the time table for LSB 3.3?
Janina- if you go to LSB’s roadmap page, they are talking about 4.0 coming out in 2009. They are thinking of major distributions of Linux.
Bill- so we have 18 - 24 months. In answer to Olaf’s question, distro’s shipping D-Bus version, that would be a 4.0 version. For 3.2 the ABI requirements for AT/SPI would not be onerous. Because of the binary requirement, we get carried over to 4.0
Bill- I propose we put the optional version aside for now and see if we can put the current version into 3.2 and look at D-Bus for the 4.0
Janina- if I understand what they said last week, they seemed to be saying no way.
Bill- this has been our intention for two years so I don’t think we should back down. Ian was negative but I don’t think Jeff was nearly as negative.
Janina- may be it is their different roles,
Bill- we need to make it clear what we are asking. Putting aside the logistics of whether it can be done, would it be reasonable to have a conformant D-Bus solution. The time to do the development
Olaf- I think LSB 4 is not supposed to be binary compatible with LSB 3. If it is a different point in time problem for clean migration.
Bill- if D-Bus were shipped, it still could be compliant, carry the old libraries until LSB 4.0.
Janina- are any distro’s on the call today?
Bill- the issue worrying Ian might not apply to us.
Olaf- if there is no defined time frame for LSB 4, people would rather do LSB 3.4 and 3.5. Another thing, generally API interfaces in the LSB must be around, that is a different suggestion. People are having different opinions about what the binary requirements. Would have a module for AT/SPI, distributions that ship qt4 are required to have it. All distributions that include CORBA AT/SPI would have to conform. In addition to having ABI present, might make
Janina- I found the url for the LSB road map and will email it. May 2007
Bill- any other comments? Do we want to ask them some other questions?
Janina- this is the time to development. We need to show up at the Berlin meeting Dec. 2-5 We could meet there people there for discussion.
Olaf- I could go but I need at home for the evening of Dec. 4. But I’m having a different opinion so it may be better for someone else to go.
Janina- our first step is to have questions. Given the small amount of lead time to the meeting, There needs to be a discussion and enough of them. We got a lot done in a week in Honolulu.
Olaf- I can be there, just need to know if it is the 4 or 5. Maybe I could ask them to move accessibility to Tuesday so I could be sure to be there.
Janina- most important is to put the questions in writing to forward to Ian and Jeff.
Bill- I could be there those days but I don’t think I could get any funding to travel.
Janina- George also thought it would be hard to get funding.
Bill- if we can achieve this as a module in 3.2 without a face to face meeting, maybe that is a tradeoff we can make. If everyone thinks it is a good idea, we should go for it. We have existing technology, You cannot standardize on something that does not exist. That is a challenge to bring on another means of conforming. Multiple paths, makes our real and gives a powerful motivation for the people who want an alternative to make it happen. Without this motivation -may take 3 years.
Janina- from a marketing stand point the time is now.
Bill- if we wait
Janina- from a real , the estimate of time to make the D-Bus version is a guess.
Gary- what is the basic motivation for having AT/SPI in LSB, is it to attract commercial AT or are we trying to legitimize the Linux for developers.
Janina-both and others. Having it in a core is a strong message. If we can succeed at it, we have the opportunity is having a compatible product that when Microsoft releases their new product. It may not be the technology we want but it is what we have and it works.
Olaf- might the fine print be changed later, if an LSB module is possible. If FSG announces separately from LSB, from political side may be easier.
Olaf- they are releasing a number of, the FSG cannot force distributions to do things
Janina- what Ian was saying, the value of LSB is the guaranty of compatibility. ISPs want see that because it gives them confidence that the products will work.
Bill- having an optional module would be useful as a degree of commitment to accessible. Unix access is still inferior to Windows, it is important to show our platform is committed to access. What fraction of that value would we get from an optional module vs. the full version. If we could show the roadmap toward having a required module.
Janina- I have convinced myself that it is not a bad option. I need to see what other products have done. I wish they had a better name than optional, like targeted.
Olaf- if qt4 was part of optional module, all the modern stuff. AT-SPI as a future direction module would.
Janina- I bet we would get excellent cooperation on this.