We all know that Linux rocks, and that the development is very
very fast. There are however some areas where core developer
disagreement, or straight out "that isn't going in the kernel
ever" stuff happens. Some good, some bad. An example is PCMCIA.
It was always an add-on, but now is integrated. Not all such
drivers are as lucky however.
Some developers just never submit a patch for something cool,
which is a reason why certain things never get put in, that are
otherwise fantastic.
Also, there are commercial companies that code exceptional things
for Linux such as VMWARE which comes with its own kernel modules.
Here is where my point comes... Some stuff in Linux like ISDN,
USB, PCMCIA have been a LONG TIME coming to get into the OFFICIAL
kernel. As Linux gains in popularity, and usage, people are more
likely to add on external patches themselves, or use a premade
kernel such as the RedHat kernel which includes various patches
by default. Also, due to the exceptional quality of products
such as VMWARE, people are more likely to be using such a kernel
add-on module.
Why does this matter? Because the more time goes on, and things
aren't in the kernel-proper, the more external people will
integrate things themselves, and the less people are using the
official kernel pure. When a "crash" occurs, the bug report is
USELESS because you are not using the official kernel, but are
using VMWARE modules, or joeblow's MM patches, and Fred Blow's
SUPERMEGADMA patches - beta versions at that.
I'm concerned that due to certain realities of people wanting to
use things like VMWARE because no alternative exists that their
reliance on that software makes their problems unheard. I
understand the why's, and agree that you guys can't chase down
bugs that may be caused by closed source software. My point
however is that like it or not, VMWARE is here, and is going to
be used by more and more people. More commercial software will
come along too of which no existing "free open source"
replacement is available. The more that people, companies, etc..
use, need, and in fact REQUIRE these things into standard
software practice, the less people will have valid bug reports to
bring forth when something goes wrong - which very likely could
have been a legit kernel bug that had nothing to do with any
addon stuff.
My concern isn't for "my case", or "company b", in which case the
common response would be "If they use a proprietary product, and
it crashes the kernel we don't care, it is there problem." I
agree with that in context of a developer, however in terms of
Linux and the real world, we could end up seeing proprietary
solutions of which no Linux free equivalent exists in effect
create a large demand for the proprietary extension or whatever,
and in fact lessen the number of people using "pure" kernels
dramatically, and thus lower the numbers of potential people who
would otherwise fit the "debugging is parallelizeable" paradigm.
I seriously think that this could be a problem in 1-2 years with
the current acceptance of Linux, and VAR solutions becoming
available like VMware. The only suggestion I have as a remedy,
is to advocate free solutions and try to promote them as much as
possible. Freemware (sp?) for example is the equiv to VMWare,
however it is not an existing useable product, so it is not worth
discussion. However, by promoting such efforts, and gaining
interest to help such a project, we could lessen the impact on
"commercial takeover" of modules and userland.
Lets face it, the kernel is the small part of the picture. If
company X dominates userland applications, then most people will
be using company X's proprietary software of which no free equiv
is available in the foreseeable future. Then company X is also
free to "fork the tree" so to speak and do damage to all the
gains we have made in the last few years.
I do believe that some companies out there do have the power to
fork Linux and get away with it too, but i think that they'd do
it in a quiet way, rather than a visible way.
Microsoft, IBM, and some other major players could all do this
IMHO. And I think some sort of discussion would be a good idea
(probably on another list, or via email).
My main reason for bringing this up, is that I am hooked on
VMware, and there simply is no replacement for it available
period. Thus any problems I have are mine. ;o( When discussing
Linux with people over the last while, I had non-linux people
saying that they want to try it out, because they heard of this
program called VMware that lets you run windows, etc.. and "my
company is installing VMware at work," etc..
Due to just the popularity of Windows itself, VMware stands a
damned good chance at getting installed on a LOT of boxes out
there, and at the same time lowering the valid number of
acceptable "Oops" reports that will get looked at here.
Could be a big problem for us if VMware really catches on... And
I can't help myself for helping spread the darned useful thing.
;o)
I'm going to go check out the status on that freemware thing now,
and cross my fingers. ;o)
Sorry for the long winded semi-OT posting.
TTYL
-- Mike A. Harris Linux advocate Computer Consultant GNU advocate Capslock Consulting Open Source advocate[insert witty random tagline or cool URL here]
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/