[lsb-discuss] What id driving the current ISO hype?

Theodore Ts'o tytso at mit.edu
Sat Aug 4 21:21:44 UTC 2012


On Fri, Aug 03, 2012 at 02:34:46PM -0400, Robert Schweikert wrote:
> >As a user of the LSB I consider it to be a standard of lower value
> >*specifically* because it's not an ISO standard, why? It displays
> >what appears to be a reluctance on the part of the LSB community
> >to work with other countries to resolve issues that are relevant
> >to all Linux stakeholders (consumers, industry and government),
> >and to work on consensus building.
> 
> Wow, that's a pretty strong statement. However, in a certain light
> you are correct. The LSB is not a consensus building Standard, and
> was never intended as such. The LSB is a trailing specification that
> documents common practice across Linux distributions, with a bent
> towards the "enterprise distributions".
> 
> If we were to "prescribe by consensus" what distributions have to
> implement we'd be out of the picture very quickly as distribution
> vendors would turn their backs if there were any perception of the
> LSB trying to prescribe what a Linux distribution should look like.

Well... there are some very gentle nudges we can apply via the open
source process if there are backwards compatibility problems in some
library's ABI (I'm looking at you libstdc++), that were introduced
accidentally.  But that's the sort of thing where we have to intervene
before software is ever released, and in most cases, before we've
tried to standardize that library.  And that's because, as Robert has
very correctly stated, the LSB is a trailing specification.

That's ultimately at the end of the day my biggest problem with the
ISO process.  If some troublemakers at various National Bodies has
some consensus which decides to require the LSB to require POSIX
compliance (and there are some good reasons by ISO-head standards
people for not wanting to have two international standards which are
almost, but not quite, completely aligned), the Linux community has a
very high probability of simply ignoring the ISO standard --- just as
various differences between the Linux run-time environment and POSIX
have been deliberately ignored for ***decades***.

Way back when, when we had the active help of various multinational
corporations, it was possible to control the outcome via very careful
placement of engineers working for those multinational corporations in
the proper National Bodies to make sure that the outcome would respect
the decisions already made by the Linux development community.

But #1, this is exactly the same thing Microsoft did when it
controlled the approval OOXML standard, and you can decide for
yourself whether the ability to manipulate the ISO standards process
in this fashion is a good thing or a bad thing --- and #2, there is no
longer a business case for those multinationals to spend $$$ to make
sure the outcome can be controlled in such a fashion so we can
guarantee a result which is friendly and defers to the Linux
development community.

So the nightmare scenario for us is that ISO decides to "add value" by
either making changes to the ISO version of the LSB standard to force
alignment with Posix, based on the consensus of the National Bodies,
or to try to hold the LSB workgroup with blackmail by voting to
approve the renewal of the LSB's ISO status only if we make certain
changes based on the the requests of certain National Bodies.

But honestly, I don't think it's that much of a nightmare, given that
the Linux devleopment community has ignored one ISO standard, it's not
that hard for them to ignore another one.  And if we publish something
which is not in sync with the ISO standard --- or rather, if ISO tries
to publish which is not in sync with what the LSB workgroup publishes,
I'm quite certain that the LSB workgroup can win the social media war
with ISO-heads....  in fact, if we do have to engage in a social media
war with ISO, it might actually enhance the LSB's credibility.  :-)

						- Ted


More information about the lsb-discuss mailing list