Re: [cobalt-developers] nice knowing you all!

[Dennis <dennis@xxxxxxxxx>]

At 12:39 PM 09/23/2000 -0800, Arthur Corliss wrote:
>On Sat, 23 Sep 2000, Dennis wrote:
>
>> Can you spell "B S D"?
>> 
>> In counterpoint, they could have used another "free OS" without the
>> restrictive GPL license (BSD comes to mind) and built an application driven
>> appliance rather than trying to generically support LINUX on their box.
>> Their boxes are nothing special (slow and expensive compared to a 1U PC
>> alternative), and they are consumed dealing with all of the pitfalls of
>> linux, keeping up with seemingly daily changes that have little to do with
>> their product, and dealing with thousands of penguinites who demand
>> unending support for free...
>
>Ahh, here we go into license bigotry.  I use both, and both have their
>strengths, no doubt.  But what spurred this discussion on was two simple
facts
>that BSD doesn't address:
>
>	--Linux was definitely a factor in the success of Cobalt.  Nice
>	  loyalty showing to consider dumping them now
>	--There are more packages ready to install for Linux than any
>	  other OS, so if you want to let them tinker with the internals,
>	  why make them compile from source?  And before you tell me some
>	  crap like that isn't an issue, you need to explore the number of
>	  neat little proggies that some nitwit developed solely on Linux
>	  without an ounce of consideration for making it portable.  As a full
>	  time IRIX user, I run into this all the time.  Those of us writing
>	  portable code (and conscious of endian and register size issues) are
>	  in the clear minority.

the original point of this thread, was the its unlikely that the
"tinkerers" are buys rather expensive raqs. most people who buy a raq

>
>Yes, FreeBSD has a Linux compatibility layer, and Solaris has one in
>development.  Do I have to tell you what I think of compatibility layers?
>Waste the hardware, go ahead, waste cycles remapping kernel calls.

yes, but freebsd ethenet driver dont lock up under heavy load, and most of
their "features" areent experimental for 10 years.
>
>> Supporting Linux proper is a nightmare, and Im sure that a very small
>> percentage of sales are to people who care...mainwhile the small pct of
>> those that do are eating their lunch in support time.
>
>That's completely ludicrous.  As a company that has to support a wide variety
>of Unices in businesses here, Linux is no more a support hardship than any
>other Unix.  In fact, with the tremendous amount of attention it gets, I'd
>wager that I find solutions quicker and easier than with any commercial Unix.

The problem with linux is that there are no solutions for its inner design
flaws. We support linux and freebsd, and believe me, linux is a nightmare.
Plus, every major release requires a major overhaul of drivers because
there is no conherent underlying structure to linux. Porting from Freebsd
3.x to 4.x, for example was a 1 day project because BSD internals are
well-defined and stable. Porting from 2.2 to 2.4 in linux will be a drawn
out nighmare because 1) 2.4 will be released with tons of poorly tested
features and 2) the basic driver structure has changed.

The fact that there are so many linux distributions makes support difficult
because they all use different libraries, so distributing binaries is a
nightmare.
>
>> Without the "hype", no commercial vendor would use linux, because its not
>> as good as BSD (an opinion shared by most people with substantial
>> experience with both) and its license is not friendly to commercial
vendors.
>
>BSD is rock solid, there's no doubt.  But the simple market reality is that
>Linux is getting the attention, and more development is being concentrated on
>it.  Applications drive acceptance, and applications are what the users want.
>In that view, moving away from Linux makes little sense.

Most of the important applications run on both linux and freebsd. Most of
the other garbage (like drivers for obscure hardware) dont work well enough
to suit commercial use anyway, so there are of little use except to the
hobbyist.

DB
>
>Your license quibble, BTW, is utter crap.  You can license any code that runs
>on top of Linux any way you want.  As long as you're not linking against
GPL'd
>code (thought much LGPL code alleviates the problem), you're fine.  And guess
>what, what application are most people using on the Cobalts?  Is it Apache?
>Is that GPL?  I thought not.  Give it a rest.

Your problem is that your ideas are trivial. 

You can't add functionality to the kernel with giving away the source,
which is why no commercial company will invest time and resources in kernel
changes (which is why the linux kernel is such garbage). So, if a
commercial vendor wanted to develop ATM (and didnt feel like waiting for
the scads of linux weenies to get it to work), they'd have to make the
source available when they are done. So guess what, they dont do it.

if you are going to spend hundreds of man-hours building a custom product,
you do it with BSD because you own it when you are done. I know of many
commercial routers based on freebsd, none on linux (and Im talking about
ascend GRFs, real routers, not some garbage based on linux that some guy
builds in his garage)

>
>If, however, you *want* to take advantage of GPL'd code, then suck it up.
>You're using their tools, you have to comply with their licenses.  Ironic
that
>you're whining about it, thought, when even the BSD's use a great deal of GNU
>cli tools.  What's that all about?  Hell, even Solaris and IRIX have GNU
>bundles ready to install off the installation disks.

Im not whining, Im just stating that GPL is anti-commercial and promote
mediocrity. Why is BSD rock solid? Because commercial vendors contribute to
making it rock solid. so they can use it as platforms for custom products.
Its a communist concept, and we all know why communism failed; Theres no
incentive to be anything by mediocre.
>
>Your political leanings in the license arena is a herring, pure and simple.

Thats right, im a commercial vendor and the GPL license inhibits our
ability use linux (not that i would want to) Your arguments are mostly
trivial and wrong. You dont have much insight into what commercial
developers consider when designing a product. The fact that very few, if
any, commercial vendors contribute to the linux kernel is a pretty good
indicator without having to argue.

DB
>
>	--Arthur Corliss
>	  Bolverk's Lair -- http://www.odinicfoundation.org/arthur/
>	  "Live Free or Die, the Only Way to Live" -- NH State Motto
>
>
>
>_______________________________________________
>cobalt-developers mailing list
>cobalt-developers@xxxxxxxxxxxxxxx
>http://list.cobalt.com/mailman/listinfo/cobalt-developers
>