From: System Manager <sys...@readysoft.es>
Subject: Linux support denial in commercial products?
Date: 1996/08/13
Message-ID: <3210ADD1.7B05@travelnet.es>
X-Deja-AN: 174362353
sender: owner-linux-ker...@vger.rutgers.edu
content-type: multipart/mixed; boundary="------------3636771D2D47"
x-hdr-sender: sys...@readysoft.es
mime-version: 1.0
x-env-sender: owner-linux-kernel-outgo...@vger.rutgers.edu
newsgroups: linux.dev.kernel
Hi all. Sorry for the long message but I think it's worth to think about
this...
To my surprise I've found that many commercial apps developers refuse to
develop Linux versions of their products. Who and why? Here's the list
I've built so far:
------> GNN & AOL
-there are versions of the GNN Server for Linux running fine. They
incorporate a pthreads library of their own
-I attach a message explaining why they're no longer supporting their
Linux port. A problem related to kernel threads questioning clone
fairness
------> Ilustra database
-they say they'll never support Linux. Too many problems with changing
kernel sources. Not enough stability for them.
------> Excite
-They had announced a Linux port of their search engine. They've changed
their opinion and there won't be a Linux version
------> Netscape
-Thay also had suported Linux since the start. One of their main
arguments to beat Explorer is that they are multiplatform. Linux is not
supported in their last version.
Not to mention products like Oracle and Informix that would be nice to
have at least the opportunity to choose.
I'm pretty sure that there are more, but I've not had the need for them.
Why is this happening? Any logical reason?
If we want Linux to become an even better alternative to commercial OS
we need to be able to compete also in commercial apps land. How can we
pretend to build business applications if we can't run commercial
databases? Possible but more difficult.
IMHO we should take actions to enforce that people to port their
products. But I don't mean putting pressure on them, but listen to their
arguments and give them answers. That can only benefit us.
My first-choice OS is Linux and I don't like not being able to run some
apps. I know the sentence: "Let's build the best database and make it
GPL'd". I'm also for that, but it should take less effort and would be
commercially better to start running those products.
Comments?
Pau
TravelNet System Manager
>> At 05:39 AM 8/9/96 -0400, Antoni Pamies wrote:
>>
>> >P.S.: When will 2.1 be out for Linux ?
>>
>>
>> There are presently no plans for a 2.1 port to Linux, FreeBSD, or SunOS
>> (that's SunOS 4.1.x, not Solaris which is supported). These three ports
>> rely on a third party beta implementation of Posix threads which is quite
>> buggy, making the port very difficult.
>
> I noticed that the GNNServer includes it's own pthreads lib and
>does not use the pthreads libs that are "built-in" to the lastest Linux
>libs. Since pthreads package is "built-in" to the lastest Linux 2.0 libs
>does that make it "third party" now? Did the integration help with
>stability? Maybe requiring a certain libs version for GNNServer will help
>out. Also, Linux 2.0 has some sort of threading built into the kernel. I
>saw something to the effect of "clone". I don't know the low-level stuff
>really well, but I though I'd ask. I work with both Solaris and Linux. I
>really like some of the features the Linux kernel provides. Good stuff.
First of all, Linux is definitely "good stuff". Nothing has done more to
bring Unix out of the universities and big business and into the hands of
ordinary people than Linux. This is good because, while I have no fear or
hatred of NT, I think competition is good and Linux is playing an important
part keeping NT from sucking up the entire low end server market.
As for the GNNserver, my understanding is that the pthreads library included
in Linux 2.0, like that used in the GNNserver, is a modified version of the
MIT pthreads library:
http://www.mit.edu:8001/people/proven/pthreads.html
There are two problems with MIT pthreads:
1. Since pthreads isn't really integrated into the system, code written
with pthreads is typically incompatible with code written with the standard
libraries and header files. This is why, for example, compiling the
postgres driver isn't a simple matter of typing make - you need to compile
against the pthreads libc and header files.
2. MIT pthreads is riddled with bugs. We had to make several important bug
fixes just to get the GNNserver to run reasonably well. A continuing
problem is that the select() system call is hopelessly broken. In 2.0 we
could get around this problem but 2.1 uses select() even more.
When we did the 2.0 port we had hopes the pthreads bugs would be fixed and
the library properly integrated into Linux in time for the 2.1 release. I
even sent a long list of bug fixes to the author of pthreads.
Unfortunately, I received no response and, as far as I can tell, the
pthreads effort at MIT has all but stopped. For example, the "late breaking
news" page above hasn't been modified since November of 1995.
> Will there be a GNNServer for Linux 2.0 when ptheads gets more
>stable?
Perhaps, but Linux will have to come a long way to convince me it's capable
of a true multithreaded application like the GNNserver. I did look at the
kernel sources to the "clone" system call. Clone *can* be the basis of
kernel thread support in Linux. In fact, it is very similar to the SGI IRIX
"sproc" system call which we use for the Irix port. However, the clone call
itself is insufficient for a complete pthreads library. At a minimum you
need mutexes and condition variables - perhaps easy to implement and
something the GNNserver development team could do - but this would siphon
off resources from more important and interesting problems (for example,
Java support, an Oracle driver, ISAPI, etc.). Basically, we wanted the
Linux port to work as well but right now I just can't justify the level of
effort required.
What I am doing is strongly suggesting all Linux and FreeBSD users upgrade
to Intel Solaris. Intel Solaris is a fully supported platform which
includes an excellent, kernel based threads library. In fact, SunSoft
reaffirmed it's commitment to the Intel platform just yesterday (see press
release below).
Reminder: Now is a great time to buy Intel Solaris. Just $995 for Solaris
2.5 plus the Workshop development environment. You can also try Intel
Solaris for free. Check out:
http://www.sun.com/sunsoft/Products/x86.html
Cheers, -Jim
_________________________________________________
Jim Davidson | ji...@sb.aol.com
GNNserver Project Manager |
America Online Inc. |
511 Bath Street | Voice: (800) 529-9166
Santa Barbara, CA 93101 | Fax: (805) 899-4316
From: Alan Cox <a...@cymru.net>
Subject: Re: Linux support denial in commercial products?
Date: 1996/08/15
Message-ID: <199608150902.KAA13255@snowcrash.cymru.net>#1/1
X-Deja-AN: 174574262
sender: owner-linux-ker...@vger.rutgers.edu
references: <3210ADD1.7B05@travelnet.es>
content-type: text/plain; charset=US-ASCII
x-hdr-sender: a...@cymru.net
mime-version: 1.0
x-env-sender: owner-linux-kernel-outgo...@vger.rutgers.edu
newsgroups: linux.dev.kernel
> To my surprise I've found that many commercial apps developers refuse to
> develop Linux versions of their products. Who and why? Here's the list
> I've built so far:
Mostly because they work on the simple basis that
estimated cost - estimated profit < threshold
> ------> Netscape
>
> -Thay also had suported Linux since the start. One of their main
> arguments to beat Explorer is that they are multiplatform. Linux is not
> supported in their last version.
Netscape 3.0beta has Linux support. Unlike the older stuff you'll notice
they now list it on their web pages too.
> Not to mention products like Oracle and Informix that would be nice to
> have at least the opportunity to choose.
Oracle 7 runs with iBCS2. Search for "Oracle Linux" for instructions on
this.
Better still see http://www.uk.linux.org/LxCommercial.html and pick one
of the many vendors (some like Empress quite big names) of databases who
do support Linux.
> If we want Linux to become an even better alternative to commercial OS
> we need to be able to compete also in commercial apps land. How can we
> pretend to build business applications if we can't run commercial
> databases? Possible but more difficult.
You can go out now and buy complete office suites (Applix, WordPerfect,..)
Databases (Empress, Flagship, RDB, SOLID, VBASE). Support the people who
are supporting you. There is some very very fine Linux product out there
it just doesn't happen to be labelled "Oracle Inc", and on the whole its
cheaper and better supported.
Alan
From: System Manager <sys...@readysoft.es>
Subject: Re: Linux support denial in commercial products?
Date: 1996/08/16
Message-ID: <321498D6.602F@travelnet.es>
X-Deja-AN: 174628902
sender: owner-linux-a...@vger.rutgers.edu
references: <Pine.LNX.3.91.960816161916.8520C-100000@www.readysoft.es>
content-type: text/plain; charset=us-ascii
x-hdr-sender: sys...@readysoft.es
mime-version: 1.0
x-env-sender: owner-linux-apps-outgo...@vger.rutgers.edu
newsgroups: linux.dev.apps
Alan Cox wrote:
>
> > To my surprise I've found that many commercial apps developers refuse to
> > develop Linux versions of their products. Who and why? Here's the list
> > I've built so far:
>
> Mostly because they work on the simple basis that
>
> estimated cost - estimated profit < threshold
>
> > ------> Netscape
> >
> > -Thay also had suported Linux since the start. One of their main
> > arguments to beat Explorer is that they are multiplatform. Linux is not
> > supported in their last version.
>
> Netscape 3.0beta has Linux support. Unlike the older stuff you'll notice
> they now list it on their web pages too.
Yes, but in the unsupported section. That doesn't make us any good.
Take for instance that you build an Intranet server for a company. They
want to purchase the Linux Netscape Browser so that it doesn't expire.
How do explain that they don't support it? We WANT them to support
Linux. We want them to sell a Linux version. We want them to recognise
Linux as a first-class OS. We have many arguments: security updates,
Posix compliance, multi-platform OS, thousands of Internet servers using
it... do I need to convince you ;)
> > Not to mention products like Oracle and Informix that would be nice to
> > have at least the opportunity to choose.
>
> Oracle 7 runs with iBCS2. Search for "Oracle Linux" for instructions on
> this.
Not the last version.
And what about iBCS2 documentation? It isn't too encouraging... it says
it isn't a supported product, that it only runs a small amount of
programs... The only programm more or less guaranteed to run is WP.
And even if you manage to run apps on it, wouldn't you prefer to
purchase a Linux version? Sure it would behave MUCH better than iBCS2.
And you would feel much more confident. Would you invest your money in
Oracle for SCO to see if you get it running in a Linux box? I wouldn't.
But I would buy a Linux version.
I could convince lots of people to try their commercial apps if they
could use the same commercial DBs that they have for the other OS.
Asking to port to another OS and another DB is more difficult. And those
big guys using Linux would take us to the top level apps land... Imagine
Linux monitoring a Nuclear Plant... but you need to get into first and
it would be too hard if you can't even run "plain office apps".
> Better still see http://www.uk.linux.org/LxCommercial.html and pick one
> of the many vendors (some like Empress quite big names) of databases who
> do support Linux.
Ok. I have had a look at all of them and I have been testing some too.
They look all right. But I'd like to find them ALL... You can find the
listed databases for other OS too, right? We have less DBs to choose
being a BETTER OS. Doesn'it it sound contradictory?
> > If we want Linux to become an even better alternative to commercial OS
> > we need to be able to compete also in commercial apps land. How can we
> > pretend to build business applications if we can't run commercial
> > databases? Possible but more difficult.
>
> You can go out now and buy complete office suites (Applix, WordPerfect,..)
> Databases (Empress, Flagship, RDB, SOLID, VBASE). Support the people who
> are supporting you. There is some very very fine Linux product out there
> it just doesn't happen to be labelled "Oracle Inc", and on the whole its
> cheaper and better supported.
Fine. They support Linux as well as other OS. Why do most of them
support SCO, Solaris, UnixWare? I don't want a Solaris to run Oracle. I
want Oracle on Linux. I want Informix and I want to run more apps than
Sun. I'd like to see the Netscape Server. I don't want to get posts from
those development teams explaning that they don't support Linux because
it dosen't have a threads package, or that transaction atomicity cannot
be guaranteed... No more excuses. Let's not be blind and say "Ok, we
have cheaper and better supported ones". They must not be that bad when
they hold a so large market share. We have an opportunity to get into
that market ONLY if they support Linux. Otherwise we'll have to fight in
the low-end servers segment. Don't tell me that Alphas, AP-1000,
UltrSparc and those beasts are not suitable to run a bank information
system. We want to reach there too. Or are we going to stop now and say
"we're better but only we know". EVERYBODY must know we are better. I
only propose another argument to make us EVEN better.
After all the effort that's being done, why not be ambitious?
Pau
From: Dominic Binks <domi...@aethos.co.uk>
Subject: Re: Linux support denial in commercial products?
Date: 1996/08/15
Message-ID: <Pine.LNX.3.91.960815092709.3413B-100000@patmos.aethos.co.uk>
X-Deja-AN: 174601143
sender: owner-linux-ker...@vger.rutgers.edu
references: <3210ADD1.7B05@travelnet.es>
x-sender: domi...@patmos.aethos.co.uk
content-type: TEXT/PLAIN; charset=US-ASCII
x-hdr-sender: domi...@aethos.co.uk
mime-version: 1.0
x-env-sender: owner-linux-kernel-outgo...@vger.rutgers.edu
newsgroups: linux.dev.kernel
On Tue, 13 Aug 1996, System Manager wrote:
> Hi all. Sorry for the long message but I think it's worth to think about
> this...
>
> To my surprise I've found that many commercial apps developers refuse to
> develop Linux versions of their products. Who and why? Here's the list
> I've built so far:
>
> ------> Netscape
>
> -Thay also had suported Linux since the start. One of their main
> arguments to beat Explorer is that they are multiplatform. Linux is not
> supported in their last version.
>
Presumably you mean you can't *buy* Netscape for Linux. I downloaded
Netscape 3.0b6 Gold for Linux, just a couple of days ago. I don't
think that they've stopped producing binaries for Linux, they are just
as is and they won't explicitly fix problems in Linux, unless they may
be problems in licenced products. The problem is that Linux is
unstable (in terms of new versions being available regularly) and not
enough revenue available from this source. Personnaly I think
Netscape are great for the fact that they even bother to produce
versions for Linux and FreeBSD despite the fact that they stand to
gain little profitwise from it.
>
> Not to mention products like Oracle and Informix that would be nice to
> have at least the opportunity to choose.
I can answer for Informix (because I know someone who has had a
discussion with Informix about using GCC on HP-UX). Informix argue
that if they allow gcc as a supported compiler and a bug is found they
really have no-one to complain to. gcc will be fixed as and when
someone gets around to it. In short Informix have no support on gcc.
In fact it shows they haven't looked into it very seriously because
I'm sure that Cygnus would support them. However it's not quite as
simple as this because Informix only support two HP C compilers on HP.
Any third party compilers are *not* supported. This is kind of
acceptable as Informix could end up supporting literally hundreds of
differnt C compilers across their platform range.
This is really quite sad as gcc is a really good choice of compiler to
support since it is cross platform. Maybe in the future things will
change.
By the way I'm saying I agree with Informix, I'm just saying this is
what they have said about gcc and that alone is good enough reason to
not port to Linux. Maybe when Caldera Open Linux is available they
may be persuaded to think again, especially as SCO Informix is known
to work with iBCS under Linux.
I would guess that Oracle is pretty similar on this.
>
> I'm pretty sure that there are more, but I've not had the need for them.
> Why is this happening? Any logical reason?
>
> If we want Linux to become an even better alternative to commercial OS
> we need to be able to compete also in commercial apps land. How can we
> pretend to build business applications if we can't run commercial
> databases? Possible but more difficult.
>
> IMHO we should take actions to enforce that people to port their
> products. But I don't mean putting pressure on them, but listen to their
> arguments and give them answers. That can only benefit us.
Sybase do provide a set of client tools for Linux but are strictly
unsupported. Perhaps if someone has a SCO Informix they could ask
Informix if it would be possible to run some of Informix's tests to
see if we can find problems in iBCS incompatibilty. If we can fix
these we may have a case to present for Informix to release an
unsupport set of tools for client side apps at least.
I use Informix as an example only because we use Informix here and I
have some experience with it.
>
> My first-choice OS is Linux and I don't like not being able to run some
> apps. I know the sentence: "Let's build the best database and make it
> GPL'd". I'm also for that, but it should take less effort and would be
> commercially better to start running those products.
Building GPL'd tools is always a good aim, but we are years behind on
RDBMS technology. Although it not take as long to get to the same
point it would take some considerable time. Anyway Postgres is a
suitable candidate to start with. But this does not take Linux into a
commercial product which is what Caldera are trying to do. Watch this
space, it may be interesting...
>
> Comments?
>
>
> Pau
> TravelNet System Manager
>
-----------
Dominic Binks domi...@aethos.co.uk
Aethos Communication Systems Ltd. | 220 Park Avenue | Bristol | BS8 1SB
Telephone: +44 1454 614455 Fax: +44 1454 620527
-----------
"Meditation, 2. _spec_ (in religious use): That kind of private devotional
exercise which consists in the continuous application of the mind to the
contemplation of some truth, mystery or object of reverence, in order that
the soul may increase in love of God and holiness of life." OED
From: Dominic Binks <domi...@aethos.co.uk>
Subject: Re: Linux support denial in commercial products?
Date: 1996/08/19
Message-ID: <Pine.LNX.3.91.960819095212.3570C-100000@patmos.aethos.co.uk>
X-Deja-AN: 175057391
sender: owner-linux-a...@vger.rutgers.edu
references: <321498D6.602F@travelnet.es>
x-sender: domi...@patmos.aethos.co.uk
content-type: TEXT/PLAIN; charset=US-ASCII
x-hdr-sender: domi...@aethos.co.uk
mime-version: 1.0
x-env-sender: owner-linux-apps-outgo...@vger.rutgers.edu
newsgroups: linux.dev.apps
On Fri, 16 Aug 1996, System Manager wrote:
> Alan Cox wrote:
> >
> > > To my surprise I've found that many commercial apps developers refuse to
> > > develop Linux versions of their products. Who and why? Here's the list
> > > I've built so far:
> >
> > Mostly because they work on the simple basis that
> >
> > estimated cost - estimated profit < threshold
> >
> > > ------> Netscape
> > >
> > > -Thay also had suported Linux since the start. One of their main
> > > arguments to beat Explorer is that they are multiplatform. Linux is not
> > > supported in their last version.
> >
> > Netscape 3.0beta has Linux support. Unlike the older stuff you'll notice
> > they now list it on their web pages too.
>
> Yes, but in the unsupported section. That doesn't make us any good.
> Take for instance that you build an Intranet server for a company. They
> want to purchase the Linux Netscape Browser so that it doesn't expire.
>From sunsite.doc.ic.ac.uk:/packages/netscape/pub/navigator/2.02/unix
netscape-v202-export.i486-unknown-linux.tar.gz
This version doesn't expire. It only has 40 bit security, but then
for international applications that's the best you'll get anyway. In
addition Caldera sell a supported version.
> How do explain that they don't support it? We WANT them to support
> Linux. We want them to sell a Linux version. We want them to recognise
> Linux as a first-class OS. We have many arguments: security updates,
> Posix compliance, multi-platform OS, thousands of Internet servers using
> it... do I need to convince you ;)
Posix compliance? I know of only one variant that is actually Posix
compliant nearly all Linux are close to Posix. The only
multi-platform support that is relatively stable current is Alpha,
although Sparc is clearly moving quickly towards stable. Linux also
has a problem with far too many configurations to be realistically
supportable. For example Slackware is mostly BSDish is file system
layout, particular /etc. Caldera is SVR4 layout. This makes support
a nightmare. The best outlook for Linux is Caldera's Open Linux which
will place Linux in the same frame as other Unix systems. It will be
Unix but just much cheaper.
>
> > > Not to mention products like Oracle and Informix that would be nice to
> > > have at least the opportunity to choose.
> >
> > Oracle 7 runs with iBCS2. Search for "Oracle Linux" for instructions on
> > this.
>
> Not the last version.
> And what about iBCS2 documentation? It isn't too encouraging... it says
> it isn't a supported product, that it only runs a small amount of
> programs... The only programm more or less guaranteed to run is WP.
>
> And even if you manage to run apps on it, wouldn't you prefer to
> purchase a Linux version? Sure it would behave MUCH better than iBCS2.
> And you would feel much more confident. Would you invest your money in
> Oracle for SCO to see if you get it running in a Linux box? I wouldn't.
> But I would buy a Linux version.
> I could convince lots of people to try their commercial apps if they
> could use the same commercial DBs that they have for the other OS.
> Asking to port to another OS and another DB is more difficult.
Yes, but this is the nub. No big DB company is going to port to a
platform that it does not believe will bring it any revenue. Linux is
perceived like that. No company using DBs is likely to port to
another DB uneecessarily and that's the status quo. What is more
likely to put the proverbial cat among the pigeons is a good Office
suite that is compatible with MS Office and UNIX equivalents,
multiplatform and requires modest resources on Linux, in comparison to
Windows systems. If in addition the solution proved to be relatively
cheap there would be a good argument for putting in our office. We
use Linux, Solaris, SunOS, Win95 and HP-UX. Currently we use Word 6/7
for our internal documentation needs and Frame Maker for some external
apps.
> And those
> big guys using Linux would take us to the top level apps land... Imagine
> Linux monitoring a Nuclear Plant... but you need to get into first and
> it would be too hard if you can't even run "plain office apps".
Personally I rather never imagine any computer monitoring a Nuclear
plant as the risks are simply too high. Unfortunately as its a
reality I have to face but not without significant concerns.
>
> > Better still see http://www.uk.linux.org/LxCommercial.html and pick one
> > of the many vendors (some like Empress quite big names) of databases who
> > do support Linux.
>
> Ok. I have had a look at all of them and I have been testing some too.
> They look all right. But I'd like to find them ALL... You can find the
> listed databases for other OS too, right? We have less DBs to choose
> being a BETTER OS. Doesn'it it sound contradictory?
By what qualification is Linux BETTER. It is cheaper certainly. It
is pretty stable on the surface and is a good toy.
>
> > > If we want Linux to become an even better alternative to commercial OS
> > > we need to be able to compete also in commercial apps land. How can we
> > > pretend to build business applications if we can't run commercial
> > > databases? Possible but more difficult.
> >
> > You can go out now and buy complete office suites (Applix, WordPerfect,..)
> > Databases (Empress, Flagship, RDB, SOLID, VBASE). Support the people who
> > are supporting you. There is some very very fine Linux product out there
> > it just doesn't happen to be labelled "Oracle Inc", and on the whole its
> > cheaper and better supported.
>
> Fine. They support Linux as well as other OS. Why do most of them
> support SCO, Solaris, UnixWare?
Because they have to make money. That's how people are employed.
> I don't want a Solaris to run Oracle. I
> want Oracle on Linux. I want Informix and I want to run more apps than
> Sun. I'd like to see the Netscape Server. I don't want to get posts from
> those development teams explaning that they don't support Linux because
> it dosen't have a threads package, or that transaction atomicity cannot
> be guaranteed... No more excuses. Let's not be blind and say "Ok, we
> have cheaper and better supported ones". They must not be that bad when
> they hold a so large market share. We have an opportunity to get into
> that market ONLY if they support Linux. Otherwise we'll have to fight in
> the low-end servers segment. Don't tell me that Alphas, AP-1000,
> UltrSparc and those beasts are not suitable to run a bank information
> system. We want to reach there too. Or are we going to stop now and say
> "we're better but only we know". EVERYBODY must know we are better. I
> only propose another argument to make us EVEN better.
>
> After all the effort that's being done, why not be ambitious?
>
> Pau
I respect you ambitions. First though you have to realise some
*facts* of the industry.
1. There is snobbery in the IT business. Companies like to pay vast
sums of money when they build a product for business purposes.
2. There is real concerns over free software and support. Many
companies are attracted to the idea of free software but if it affects
their core business due to lack of support they will not choose it -
they cannot afford.
3. There is a prejudice over Linux. People do not like being told
they could get better for less so they use perceived differences are
bargaining chips.
4. Linux zealots do not add to the cause. Linux zealots get the backs
of companies up who may have considered Linux but have decided not to
procede for any number of reasons.
-----------
Dominic Binks domi...@aethos.co.uk
Aethos Communication Systems Ltd. | 220 Park Avenue | Bristol | BS8 1SB
Telephone: +44 1454 614455 Fax: +44 1454 620527
-----------
"Meditation, 2. _spec_ (in religious use): That kind of private devotional
exercise which consists in the continuous application of the mind to the
contemplation of some truth, mystery or object of reverence, in order that
the soul may increase in love of God and holiness of life." OED
From: d...@neteng.engr.sgi.com ("David S. Miller")
Subject: Re: Linux support denial in commercial products?
Date: 1996/08/19
Message-ID: <199608191001.DAA04702@neteng.engr.sgi.com>#1/1
X-Deja-AN: 175067041
distribution: local
organization: Mail -> News Gateway
reply-to: d...@sgi.com
newsgroups: linux.dev.kernel
Date: Mon, 19 Aug 1996 10:20:16 +0100 (BST)
From: Dominic Binks <domi...@aethos.co.uk>
Yes, but this is the nub. No big DB company is going to port to a
platform that it does not believe will bring it any revenue.
It may not be like this forever. There are buisnesses who use high
end DB's where performance, performance, performance is strictly the
one and only concern. Nothing else matters.
What do you think would happen if, for example, I walked into Charles
Schwab's Wall Street office tomorrow and told them "I can get a
transaction from one machine to another done XXXX percent more quickly
than you do now, that is, if use a Linux based solution." This means
Schwab would be able to cut deals before any other broker could
because they have the fastest technology.
And as Larry McVoy, myself and numerous others have shown the world.
Linux _does_ have the performance edge, so we could plant ourselves in
critical positions like this.
By what qualification is Linux BETTER. It is cheaper certainly. It
is pretty stable on the surface and is a good toy.
Linux has been shown to irrevocably have the best performance in
numerous critical ways. As everyone knows, we have the _fastest_ over
the wire TCP numbers in the world.
> I don't want a Solaris to run Oracle. I want Oracle on Linux.
You can get halfway there on a Sparc these days. We have the SunOS
Oracle working under emulation with SparcLinux. The SparcLinux team
and myself are going to work towards making sure we can get some
legitimate Oracle benchmarks run with such an installation to show the
world that we beat the pants off of SunOS/Solaris running SunOS Oracle
on a SparcLinux box.
If people are a little daunted as to how much I say that performance
can have an effect on Linux being able to get into big commercial
markets, think about this. Some companies make their entire
livelyhood around having the fastest solutions on the planet. Check
out SGI for example.
2. There is real concerns over free software and support. Many
companies are attracted to the idea of free software but if it affects
their core business due to lack of support they will not choose it -
they cannot afford.
Expect this to change. If you _really_ want Linux support at a big
shop, there is certainly someone out there even today who are willing
to get paid to do it and be that "person on the phone".
4. Linux zealots do not add to the cause. Linux zealots get the backs
of companies up who may have considered Linux but have decided not to
procede for any number of reasons.
I think one of the FreeBSD people referred to us as "Cowboys", to that
I respond "Yee Fuckin' Haw"!
d...@engr.sgi.com
'Ooohh.. "FreeBSD is faster over loopback, when compared to
Linux over the wire". Film at 11.' -Linus