Re: cooperation with ISO

On Tue, Mar 25, 2003 at 11:38:29AM -0800, [email protected] wrote:
> The closest well known formal standards body to open source/free software 
> in spirit is the IETF: but they generally try not to standardize API's 
> at all (the socket API is about the only thing they've done). I'm intimately 
> aware of the IETF process, the people and the culture, having been the 
> editor of the HTTP/1.1 specification.  During this period, I reported 
> into the standards part of Digital/Compaq and found it educational: from 
> a vendor's point of view, vendors often go "standards body shopping" to 
> find one they believe they can manipulate via corporate or country voting 
> to get the outcome most favorable to them.  The companies motives varied 
> from very honorable to very dishonorable, depending on the company and 
> circumstance.

Yes, I agree that we really should try to do an open organisation, with
membership based on individuals. I have myself been working a lot with
standards, including being an editor or coeditor of 2 RFC and very much
involved in some of the other RFCs - especially MIME and ESMTP. 
I agree that IETF is close to what I want for X, but there are some
other possibilities. My suggestion on using ISO was mostly because
thay have started up this new study group on Linux with a meeting in
June.

Also my experiences with the openness in different standardization
organizations tells me that ISO is not the worst place for open
standardization, where individuals are counted in. I have quite some
experience with standardization, having been the editor of 6 ISO
standards or technical reports, 2 RFCs, 1 IEEE standard and about 5
specifications in the European standards organization CEN, and then
participated in a lot of standardization activities in a number of
other standardization efforts in said ogranizations and other places.

My test for if a standards organization is an open organisation, is
whether people that do not represent big commercial companies regularily
have leading roles in the organization, such as editor or chair.
I am myself just representing a 1-man company, or a user group,
the Danish Unix Users Group (www.dkuug.dk).

I regularily see 1-man company persons being a leader in the following
organizations: ISO, CEN, IETF and IEEE. I have not really seen it
often on other standardization organizations.

Yes, it is true that companies go standards shopping. And I would
really like us to avoid X standardization being dominated by companies.
I feel that the trend is that the corporate side is trying to take over
the Linux business, that was created by thousands of volunteers all over
the world, in constortia and organizations that are set up and run by
the big Linux vendors. I would really like if X was not going down that
route. Another trend I see is that a lot of Linux steering is coming out
of the USA, especially Silicon Valley, and I would also really like to
see an organization for X that is truely international.

> X.org does not currently have any mechanism for individual participation. 
> This is a fundamental bug of most standards bodies (not of the IETF), 
> and unless that were fixed, is a fundamental non-starter, in my view.  
> To fix it, if I remember correctly, would require renegotiation of all 
> contracts of the members, and most of the members who are participating
> have been coming from the commercial UNIX side of their respective
> companies.  Company based voting on standards with no voice from
> individuals not affiliated with them gets us nowhere, and gets us
> there very slowly, I think.

IEEE also has individual membership. IEEE is the one that standardized
POSIX, and thus very close to X. And they do standardize APIs.
I think IEEE could be a good candidate for X to get some specifications
done as formalized standards.

> This leaves us with a major puzzle, if we believe formal standardization 
> is needed at all (it probably is: the question is when).  My view is that 
> the rigor required by good standardization processes is very desirable 
> in the long run. The closest organization to open source's views is probably 
> the Free Standards Group, I suspect; having not worked with them, I have 
> no first hand opinion of that organization, and whether it is a viable
> venue for formal standardization.

I have worked with the Free Standards Group. My personal opinion is that
it seems to be all the big Linux vendors getting together and setting
some new Linux standards. One is of cause the LSB - Linux Standards
Base. This was made to make it feasible for 3rd-party software vendors
to create just one binary package for all Linux distributions, so of
cause that standard was dominated by Linux software vendors and Linux
distributors.  But so are also their other standards, IMHO. One example
where I was involved is internationalization, where I actually am the
editor of the new Locale specification in ISO (14652) which is more or
less implemented in glibc today. The LSB people set out to standardize
something different from what was already implemented in Linux, 
and they allocated the leading posts such as chair and editor upfront,
before community people were really let in to the organization. 
To me it seems like the Free Standards Group is led by the leading
Linux companies of California and Japan, and they created it because
they did not want to go standards shopping in ECMA, ISO or IEEE.

> Despite the above statements, I think your view of ISO, Alan, is exactly 
> correct. My view of both ISO and ECMA is pretty poor, having watched from 
> the perspective inside a big company manipulations of them by various 
> vendors, and the politics thereof.

I agree with the opinion on ECMA. For example Microsoft went standards
shopping with C# and CLI that were first ECMA standards, then
fasttracked as ISO standards. I do think that they first went to ECMA
because they thought they could control the process well there.
Although I have been involved in standardization of programming languages
for a long time, I found it hard to get involved via ECMA for these
undertakings, so they probably were right.

On the other hand I have been very involved in the ISO standardization
of C and C++. This has been a quite open process, as far as I can tell.
People from very small companies participate and have a lot of
influence. C and C++ are actually also good examples of development
in ISO of libraries. C and C++ are some of the most successful
programming languages in the market today, and I believe most of X
software is written in C or C++, and so are the GUIs that build on X.
So don't say that you cannot develop APIs in ISO.

My original thought with this tread was to prepare for the ISO meeting
in June, and I have already last August suggested to those people that X may be a
idea for ISO standardization. Maybe it is not. But then I would
like to tell people at the meeting why. I understand that a number of
people here on the list, including prominent persons, do not fancy to do
work on X in ISO. Maybe they are right. But maybe they have not seen
the different possibilities there are for the cooperation with ISO,
or they have not experienced how ISO work on libraries and operating
systems, like what is done in ISO/IEC JTC1/SC22 on C, C++, Ada, POSIX,
internationalization etc (http://www.dkuug.dk/jtc1/sc22). 

best regards
keld