[lsb-discuss] Can we find a fit for LSB and Mobile?
tytso at mit.edu
Sun Apr 15 03:55:14 UTC 2012
On Fri, Apr 13, 2012 at 02:05:38PM -0400, Robert Schweikert wrote:
> 2.) One of the core charters, at least for as long as I have been
> involved in the LSB, was to attempt to provide an environment that
> is targeted towards the Enterprise market.
I don't know that I would characterize it that way. Historically, I
think it would be fairer to say that the goal was to focus where the
LSB could make an impact. That happened to be the enterprise market,
mainly because the year of the desktop was always "next year", and so
while there were essentially *no* ISV's that were really focusing on
the desktop market (except for the enterprise desktop/workstation
market) there were ISV's selling Linux products into the ISV space.
As far as Mobile devices are concerned, my advice to the working group
is that it only makes sense to pursue it if there are engineers from
the mobile space, and/or sponsors from the mobile space willing to put
significant, cold, hard cash, to sponsor the effort. If there are
sponsors from Tizen who are willing to assign engineers and/or put in
enough cash on the table so that the ABI for QT5 can be documented,
added to the LSB database, and tests written for Qt5, then great!
This will save work when and if the enterprise distributions start
picking up Qt5, and enterprise ISV's start wanting to use Qt5.
However, it's probably not fair to ask Robert to work on Qt5, since
it's probably not relevant for his employer at the moment.
This is somewhat more complicated for the people who are paid by the
Linux Foundation who provide central infrastructure services, since
ultimately their salaries are paid by the Linux Foundation's sponsors,
and it's ultimately Jim Zemlin's call whether or not work needed to
support new initiative, such as Mobile, is likely to add/continue the
sponsors being willing to fund the Linux Foundation. To the extent
that machines and salaries are being paid for by the Linux Foundation,
this is not a trivial thing.
One of the reasons why I pushed for an Java module was because it
would have significantly increased the likelihood that a Platinum
sponsor would continue its support for multiple years into the future.
As it turns out Sun wasn't willing to pay ball, so we had to drop that
plan (although if they had been willing to play ball, it might have
reduced the chances that Oracle could have claimed that it's possible
for a company to copyright a computer language; we'll never know how
the what if's had played out.)
However, I'd use this as an argument to strongly consider diversifying
what LSB stands for beyond enterprise, because I have strong doubt
that the traditional sponsors who have been interested in eterprise
will continue to support it strongly. So having additional industry
markets which the LSB can support, in order to broaden the base of
both volunteer and corporate resources supporting the LSB, is a Good
Thing. Ultimately that will also help people who are only interested
in the LSB's applicability to Enterprise, because they won't have to
be the only people supporting the LSB.
Whether or not Mobile is the right area I think remains to be seen.
For better or for worse LSB isn't a good fit for Android, since the
bulk of Android applications aren't written to the traditional LSB
interfaces. Whether or not the LSB is a good match for Tizen is I
think still an open question. I will point out it will only be a good
match if there is extremely strong support from the Tizen community,
both from the distro and ISV side of the equation --- and that support
has to include both monetary and engineering components.
And if the people who are pushing "LSB on mobile" weren't thinking of
Tizen, we should be clear exactly what specific mobile platform they
are thinking of. Because ultimately that's very, very important.
> Anyway, to sum things up. IMHO LSB and Mobile do not mix very well
> by definition. This is not to say there cannot be other compliance
> programs that use some of the work we do in LSB, but these should
> not be called LSB. Also, the whole thing really has nothing to do
> with modularization.
So here I would urge folks to consider very strongly what is meant by
"LSB". Do you mean the LSB certification trademark, which is owned by
the Linux Foundation? Do you mean the resources (both people and
servers) which are paid for by the Linux Foundation? Do you mean "the
LSB development community"? Or do you mean something else?
Back when the sponsors of Meego were interested in using the LSB (and
which by the way funded a lot of work done by ISPRAS that was also of
use and interest to LSB in the enterprise world), they were not at all
interested in using the LSB certification mark, but rather the testing
and certification processes and software that was done by the LSB. So
it was not "called" LSB, and perhaps not everyone was aware that some
of the "LSB work" was being funded by people who were ultimately only
interested in Meego. But to the extent that this broadened the
support base of the LSB work, it's useful, and if modularization means
that more people are willing to fund work of the LSB, then I would
argue that it very much does matter.
More information about the lsb-discuss