On Sun, 18 Feb 2001 [EMAIL PROTECTED] wrote:
> On 19 Feb, Rev Simon Rumble scribbled:
> -> On Sun, Feb 18, 2001 at 04:19:07PM -0800, [EMAIL PROTECTED] uttered:
> ->
> -> > they want to release a product. they producs requires feature that is
> -> > beta and in development in version X of the kernel (or library or app)
> -> > for theirs to function - they cannot release until their dependancies
> -> > have been released - thus their cannot set schedules.
> ->
> -> If it's that critical then they need to put their own resources into
> -> making it happen. Easy. Doesn't have to be outside the mainstream
> -> development, just them hacking on _their_ problems and getting things
> -> release-ready.
>
> but if the feaure they want is done - but the rest of the subsystem
> they depend on still needs work according to the developer before its
> released - they're stuffed. what if the developer decides to go on a
> month's world tour and not do anything? he is not beholden to do any
> release - regardless of how much work the company does. thy are stuffed
> waiting on someone who has no guaranteed schedule and who they cannot
> nudge into having a schedule.
>
> sofar in the linux world this has resulted in the company hiring the
> person they depend on in the end if they can afford it - but some
> cxompanies cannot, or the person just wont move and wants to stay where
> they are - then the company is USCWAP
OK, say I'm a company developing a product in parallel for Windows and
Linux. I don't really have control over either, I can look at the stated
schedules, but has windows ever been delivered on schedule? Or I can
work based on what's available right now, again, equally valid for both
Windows and Linux development. Or I can try to influence the development
of the product, in the Windows case with a contract of some kind, and in
the Linux case by paying the person responsible. Either of these
approaches could fall through for all sorts of reasons, but in the Linux
case the developer isn't likely to change the system to actively screw
you over because they have a commercial interest in a competing
product. Agains, possible problems both ways, but i'd say you are better
of developing under Linux.
So what's the solution? Of course it's to bring critical functionality
in house, even if that means you duplicate work that is being done
externally. This is what happened with Nvidia, for example, producing
inhouse kernel modules for driver support. If you can't afford to do
this, you are possibly screwed, and it doesn't matter whether you are
developing for Linux or Windows. I don't see where there's a substantial
criticism of linux in this point. You can also easily be USCWAP
developing for Windows.
> -> > IMHO there should be a non profit organisation thatis funded by all
> -> > companies using linux that people work for (probably the top developers
> -> > most likely) who just ork on improving things and rolling back changes
> -> > the commercial vendors make intoa central distribution of software
> -> > compnents so its kept organised.
> ->
> -> I guess this could be okay, though it needs to be managed right. It
> -> does sound like Linus' bandwidth is somewhat constrained.
>
> the problem here is that linus is a "central point of failure" - hes
> good - but as anyone knwos -a central point of failure is bad.
> corporations at least have lots of people and thus no central point.
> they can re-assign mantinence when people go on holidays, its more
> dependable for them. thats they point the article was making. they have
> a very good point.
There are "central points of failure" in every company I've ever worked
for. But you make it sound like if Linus went under a bus tomorrow
(heaven forbid) that would be the end of Linux. I don't even think it
would be a terribly large hiccup. There'd be some reshuffling, but there
also would be if key people were lost in any commercial enterprise. In
fact I think Linux would be less vulnerable (in terms of work getting
done) to a failure like this than many private companies would be. Alan
Cox would presumably fill Linus shoes as an interim measure immediately,
and work would continue. There are "lots of people" working on linux as
well, and a bit of reshuffling wouldn't be such a big problem.
cheers,
Martin
--
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://slug.org.au/lists/listinfo/slug
