From: keith@ksmith.com (Keith Smith) Subject: Re: Is this becoming comp.linux.advocacy? Date: Tue, 10 Aug 1993 22:28:22 GMT
In article <CBH3w0.3Bt@telly.on.ca> evan@telly.on.ca (Evan Leibovitch) writes:
>In the long run? A year from now, Linux will be a very different thing
>than it is now. So will Univel, SCO, and the others. While Linux will
>be guiding its basic networking and binary compatibilities through
>infancy, commercial products will (and some already do) support
>internationalization, enhanced security, NetWare, multiple processors,
>multi-threaded processes, enhanced filesystems, and Pentium-optimized
>compilers. The performance improvements alone may enough to justify the
>cost differences between Linux and commercial systems. And there's still
>no guarantee that Linux documentation will be at a level acceptable for
>commercial consumption. But that's long-term.
enhanced security: Important for BIG business and secure applications,
but this is a fairly small market.
Netware: What do you mean? SPX/IPX?
Multi-Processor & Multi-Threading: WEEELLLL, Maybe. At the present
time there are no HARDWARE MP standards other than proprietary ones. I
don't think your going to see a mass jump on the MP with Unix bandwagon
until this is corrected. Multi-Threading is neat too. It's the
application for the end user I have to ask about.
Enhanced Filesystems & Pentium optimizations: This is just plain dumb &
stupid. Linux ext2fs is _already_ faster than my SCO stuff, and so is
the kernel. The GNU compiler is the standard against which the
commercial compilers are judged. Get real. Also, how long did it take
the commercial vendors to ADOPT faster filesystems? Huh? Oh, GEE and
WHERE did they get the technology to BUILD them? Oh, Berkley, that's
right, they built from *FREE* software.
>For the foreseeable future, anyone running a business, and needing a
>multi-user system to run their business, would be a fool to consider
>Linux. None of the major database vendors (Oracle, Informix, Progress,
Then I am a fool. If Linux would run COFF binaries (which is apparently
on the horizon) I would install several TOMORROW!
>etc.) even know what Linux is, let alone have any porting plans. And
>that means that none of the thousands of vertical-market packages that
>VARs have developed for Unix (not to mention all the custom apps
>written on these 4GLs) will run on Linux. Even if Linux can develop the
>technical capability to run these apps, they won't be supported on Linux.
Well that depends on the user base. You get enough of a user base and
this too can change. In a market like these crazy computers, anyone
dumb enough to say something _WON'T_ happen ought to have their head
examined.
>*That's* why most businesses run Unix, because Unix runs their
>businesses. Talking them into trusting Linux will be a far harder
>sell than Usenet cheerleading. And anyone installing Linux at a
>client site without fully disclosing where it comes from and who
>supports it, had better take out a decent liability policy.
Oh jeez. What kind of installations are we talking here? If I do a
custom install of some horizontal custom application it's on my head
anyway. Either *I* trust Linux to do what I need or *I* don't. I mean
what is the WORST thing that can happen? The whole disk get's trashed
and I reload from backups ... ??? Of course this has _NEVER_ happened
with SCO or SVR4 ( In a pigs eye ).
If my operation was critically dependent on a SINGLE system being up I
probably wouldn't run vanilla SCO on vanilla hardware either. I'd have
some fault tolerant setup running whatever software with full software &
hardware maintenance agreements. That stuff is _STILL_ mostly
proprietary so you can't really compare it.
-- Keith Smith keith@ksmith.com 5719 Archer Rd. Digital Designs BBS 1-919-423-4216 Hope Mills, NC 28348-2201 Somewhere in the Styx of North Carolina ...