
                   comp.os.os2.beta                 (Usenet)

                  Sunday, 26-Sep-1999 to Friday, 01-Oct-1999

+----------------------------------------------------------------------------+

From: as@sci.fi                                         26-Sep-99 10:12:18
  To: All                                               26-Sep-99 18:36:19
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: Anssi Saari <as@sci.fi>

Lee Pearson <warp@ktn.net> writes:

> During startup, while a dialog is displayed with the 
> title "Gathering Information. Please, Wait!", the following
> error pops up.
> 
> Critical Error
> TList error exception (EListError)
> occured at $0001048C !
> Terminate program?
> <_Yes> <_No>
> 
> Yes results in:
>  
> Exception occured:
> List error exception (EListError)
> occured at $00038BF5
> Program is terminated.
> <OK>
> 
> No results in the program exiting silently(no other errors.)

I get the exact same thing. Warp 4, FP11 here.

-- 
Anssi Saari - as@sci.fi

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Tampere University of Technology (1:109/42)

+----------------------------------------------------------------------------+

From: daolath@news1.mnsinc.com                          27-Sep-99 02:21:27
  To: All                                               27-Sep-99 04:11:04
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: daolath@news1.mnsinc.com (Mike Williams)

On 26 Sep 1999 10:12:37 +0300, Anssi Saari <as@sci.fi> wrote:
>Lee Pearson <warp@ktn.net> writes:
>
>> Critical Error
>> TList error exception (EListError)
>> occured at $0001048C !
>
>I get the exact same thing. Warp 4, FP11 here.
>

Me too, with Warp4 FP11. Any word as to a work-around/fix?

-- 
-Mike
I can also be reached at work: mike.williams@swift.com

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Verio (1:109/42)

+----------------------------------------------------------------------------+

From: Trancser@Freenet.Nether.Net                       27-Sep-99 03:51:07
  To: All                                               27-Sep-99 10:29:22
Subj: Think IBM'll ever....

From: Trancser <Trancser@Freenet.Nether.Net>

Change either HPFS or HPFS386's code, so that it will dynamically
allocate (and release) disk cache, rather than having to specify how
much (incredibly valuable) memory/ram to lock (non-swappable) as disk
cache? I myself, cant help but to admire how nicely Linux runs with
dynamic disk cache, it really helps execution speed as well as less disk
chatter when using the same program more than once! I'm sure the same
could be done for OS/2, which would probably make it even TONS faster
than it is, presently!

This would probably just be wishful thinking (mostly on my part), but if
this what I understand to be called "MMAP" (which is currently supported
by Linux) which also sounds neat; executables are supposedly never
actually stored in swap memory, but are rather re-loaded on-demand from
the original executable on the hd (parts are not pulled back out of swap
file - which might cause more disk activity?), and only data generated
by the program are stored into swapper file. This might sound impossible
by many people (when I heard about it myself I was...and still am sorta
skeptical) but supposedly thats what happens...

Oh well, anyway....adding the dynamic disk cache option might be more of
a probability, than this MMAP thing would be probably...but at least if
had dynamic disk cache wouldnt have to worry about possibly tons of
memory being locked and unswappable...even if your specifing it (no disk
cache = constant disk activity, but having even more than a megabyte of
memory reseved as disk cache might cause problems on some systems...)!
...hm....anyone think its possible to write another hpfs driver, for
doing this? :)





--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: EarthLink Network, Inc. (1:109/42)

+----------------------------------------------------------------------------+

From: sbowring+nospam@mpc-data.co.uk                    27-Sep-99 13:53:05
  To: All                                               27-Sep-99 15:18:03
Subj: Re: Think IBM'll ever....

From: "Simon Bowring" <sbowring+nospam@mpc-data.co.uk>

On Mon, 27 Sep 1999 03:51:15 -0400, Trancser wrote:

>Change either HPFS or HPFS386's code, so that it will dynamically
>allocate (and release) disk cache, rather than having to specify how
>much (incredibly valuable) memory/ram to lock (non-swappable) as disk
>cache?

No, I'm quite sure they never will (at this late stage).

> I myself, cant help but to admire how nicely Linux runs with
>dynamic disk cache, it really helps execution speed as well as less disk
>chatter when using the same program more than once!

Linux is a much more streamlined OS than OS/2, NT, Solaris, AIX etc
which all contain more "layers" of software, so linux ought to be 
outperform them all (except in disk<->network performance, 
because the OS/2 Server + HPF386 combination, has a direct device
driver level interface between the LAN and disk subsystems, which
is very fast).

NT and possibly Linux use a dynically sized disk-cache which uses all
available (unallocated) memory for disk cache - this is fine if you have
much unallocated memory for any length of time, but the moment you become
"overcommitted" [i.e. more stuff is in (virtual) memory than will 
fit in real memory] this scheme effectively degrades to a fixed cache 
scheme ('cos you have no "unallocated" memory - you're effectively in 
a memory debt situation, "borrowing" off disk space)!

So, with such a unified memory manager and dynamic disk cache, you 
will get fastest performance when your machine has lots of free RAM 
and nothing swapped out (i.e. it would already be fast, since it has 
nothing to swap back in from disk).

As memory commitment increases, your disk I/O will fall off!

With a fixed cache, as under OS/2, memory commitment does not impact the 
average disk I/O speed, so there is less degradation in performance
as memory usage grows (so it starts off slower, but doesn't get slower
as memory commitment increases).

If I were designing a new OS, I would use a unified scheme, but this 
gives the biggest performance boosts in the situations where you least
need it (which is a bit "arse about tit")!

Traditioally most (32 bit) OS/2 systems have run well over-committed,
so the massive rewrites necessary for unified memory management would yield
little benefit to most users.

>executables are supposedly never actually stored in swap memory, but are 
>rather re-loaded on-demand from the original executable on the hd 
>(parts are not pulled back out of swap file - which might cause more 
>disk activity?), and only data generated by the program are stored 
> into swapper file. This might sound impossible by many people (when I 
>heard about it myself I was...and still am sorta skeptical) but supposedly 
>thats what happens...

OS/2 versions 1.x to 2.x always used to behave as you ask: when 
overcommitted, (least recently used) code would be discarded from 
memory, so when the code was required again, it could be recovered 
from the executable.  Trouble is, this is slow for these reasons:

1. The code loaded in has to be relinked into place  - all address
   contained into the newly re-loaded code to other bits of (already 
   loaded) code and data have to be recalculated (aka fixed-up), this
   is timeconsuming.
2. The code may reside on slow media such as CD-ROM or across a network.

So with Warp 3.0, IBM came up with the optimization of loading a 
tonnes of (system) DLLs etc at startup, but instead of being marked 
as "discardable", they were marked "swappable", since recovering 
code from the swapfile requires no fixups, and the swapfile is 
guaranteed to be on fast-ish media.

There are various patents on the sophisticated memory management and 
task scheduling algorithms in OS/2!

>its possible to write another hpfs driver, for doing this? :)
Not practiclaly possible, I wouldn't think!

BTW: I don't understand why OS/2 seems sooo sllooowww at loading
(fixups?) and especially unloading programs, anyone?

Simon Bowring



--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: MPC Data Limited (1:109/42)

+----------------------------------------------------------------------------+

From: OS2Guy@WarpCity.com                               27-Sep-99 08:32:22
  To: All                                               27-Sep-99 15:18:03
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: Tim Martin <OS2Guy@WarpCity.com>

Andreas Linde wrote:

> hello
>
> >When I try to use it, I get the
> >following errors:
>
> >1) "cannot open ini-file at $0001048C !
> >Terminate Program ?"
>
> >if I answer YES then  I get
>
> >2) "Exception occured: Cannot open
> >ini-file at. $00038bf5  Program is
> >terminated"
>
> same error here....

As with all 'new' programs for OS/2 I gave it a test
run in hopes of bringing it to the attention of Warp
City members.  It is very well done and did work
for 'awhile' but the longer it was up the more often
I found it went into a freeze.  Several times that freeze
forced me into a full reboot to recover.

It will be a handy and very professional system
information tool once the bugs are ironed out.
Once 'ready for the masses' we'll be happy to
promote it.

Tim Martin
The OS/2 Guy
Warp City
http://warpcity.com
"E-ride the wild surf to Warp City!"


--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Warp City (http://warpcity.com) (1:109/42)

+----------------------------------------------------------------------------+

From: andreas.linde@os2.org                             27-Sep-99 16:27:11
  To: All                                               27-Sep-99 15:18:03
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: "Andreas Linde" <andreas.linde@os2.org>

hello

>When I try to use it, I get the 
>following errors:
>
>1) "cannot open ini-file at $0001048C ! 
>Terminate Program ?"
>
>if I answer YES then  I egt
>
>2) "Exception occured: Cannot open 
>ini-file at. $00038bf5  Program is 
>terminated"

same error here....

ciao

 andy

+------------------------------------------------------------------+
  www.OS2.org
  Webmaster/-designer 
  Andreas Linde
  email: andreas.linde@os2.org
  irc-nick: kerni
+------------------------------------------------------------------+



--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: OS2.org (1:109/42)

+----------------------------------------------------------------------------+

From: franks@owt.com                                    27-Sep-99 08:18:07
  To: All                                               27-Sep-99 16:32:17
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: "frank schmittroth" <franks@owt.com>

On Mon, 27 Sep 1999 02:21:54 GMT, Mike Williams wrote:

>On 26 Sep 1999 10:12:37 +0300, Anssi Saari <as@sci.fi> wrote:
>>Lee Pearson <warp@ktn.net> writes:
>>
>>> Critical Error
>>> TList error exception (EListError)
>>> occured at $0001048C !
>>
>>I get the exact same thing. Warp 4, FP11 here.
>>
>
>Me too, with Warp4 FP11. Any word as to a work-around/fix?
>
>-- 
>-Mike
>I can also be reached at work: mike.williams@swift.com

I don't see the error on my Warp 4 FP11 system.
frank.


--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: http://extra.newsguy.com (1:109/42)

+----------------------------------------------------------------------------+

From: bbarclay@ca.ibm.com                               27-Sep-99 16:47:22
  To: All                                               27-Sep-99 21:33:15
Subj: Re: HPFS386, The Reality...

From: Brad BARCLAY <bbarclay@ca.ibm.com>

askbill*AT*ibm.net wrote:
> If that were all, it would have been/still would be cheap and cheerful
> enough to modify "ordinary" HPFS to use a larger cache......

	Except that the IFS model in many versions of OS/2 still uses some 16
bit addressing.  It's an addressing limit that places a 2Mb max cache
size HPFS.IFS, not something in the IFS itself.

	OS/2 WARP Server for e-business introduced a new 32 bit IFS model which
is used by JFS for using bigger cache sizes.  Theoretically, for WSEB
users, a new HPFS driver could be coded which can take advantage of the
larger possible cache sizes, but those in these groups using anything
prior to WSEB wouldn't be able to use it (and those of us with WSEB can
always use JFS for anything that benifits from big caches).
 
> IMNSHO, the re-coding to run HPFS386 in Ring 0 is actually more important,
> partly because it also applies when either a small-memory model reduces
> the ability to take advantage of larger cache, or a high-end RAID
> controller provides its own hardware caching.

	This only becomes important in servers which are fielding quite a large
number of IO requests from fast media.  On your typical client machine
you'll hit the drive and physical controllers IO limits before you'll
hit your CPU's limits in handling the IO requests.
 
> In Weendoze & EIDE, for sure.  In OS/2 it might be more accurate to say:
> 
> "...is spent by the CPU doing other useful things while waiting for a
> response from the hard drive."

	I was assuming a one application situation where the app waits for the
completion of IO before going on to whatever it needs to do next.  This
was done to simplify the argument, ubt you're right - OS/2 will happily
do other instructions for other threads while any one (or more) threads
are blocked on IO.
 
> >       For those who really want to improve their disk speed, there is no
magic
> >pill.  Go SCSI and buy a nice, quick UltraWide SCSI drive (on my Quantum
> >Atlas 4 I'm doing between 16 and 24 Mb/s, with a 6.9ms access time :).
> 
> Two "magic pills" are to use multiple smaller drives (cheap but cheerful),
> then hardware duplex (level one RAID - not so cheap).

	For client based systems I'd say this is unnecessary.  Many modern SCSI
drives have independent heads which can help take care of this for you
without requiring any added hardware, while remaining entirely
transparant to the OS itself.	
 
> Reads, OTOH, benefit from the ability to find identical data under EITHER
> of two head-assemblies/rotating media sectors, one of which will
> statistically be closer to the data than the other.

	Well, that depends.  Sectors are only read in sequential order if
you're using a FCFS (first come, first serve) buffer on your sector
requests.  Most modern drive and filesystem controllers don't use FCFS
on sector requests.  More commonly, sector requests are placed in a
priority queue which ranks the next sector to be read based on one of a
variety of possible criteria.  A step up is to use SSTF (shotest seek
time first), which enqueues incoming sector requests based upon which
ones take less effort to reach based on the current drive head
positioning.  A step up from that is the SCAN enqueuing mechanism, which
is similar to SSTF, but is unidirectional (ie: not just the closest
sector, but the closest sector in the direction the head is currently
moving).  There are variations on thi as well, incluing c-SCAN, n-step
SCAN, VSCAN and CVSCAN (and others!).  For a more detailed discusion of
these, see http://yaztromo.idirect.com/seekopt.html

	The point being "statistically closer" may not matter - indeed, it can
cause the paths to potentially lengthen.  The real benifit in the
mechanism you propose is that you can do simultanious IO - while one
drive is reading the first sector, the second drive can be reading the
second.  In this sort of situation, you wind up losing about half your
head latency on average.
 
> None of this requires HPFS386.  All of it benefits from it, especially in
> 256 MB or so of RAM and on dual-CPU's.

	Well, in the end faster interfaces and faster drives give you the real
speed increases.  Plus, anything which reduces CPU load (such as SCSI
interfaces) is also going to give you a boost.  

> Any computer which has not anticipated your next move is too slow!

	Overall, I think we're in agreement:  speed is good :).

Brad BARCLAY

=-=-=-=-=-=-=-=-=-=
Posted from the OS/2 WARP v4.5 desktop of Brad BARCLAY.
E-Mail:  bbarclay@ca.ibm.com		Location:  2G43D@Torolabs

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: IBM Toronto Labs, DB2 for OS/2 Install Developer (1:109/42)

+----------------------------------------------------------------------------+

From: donm@ftel.net                                     28-Sep-99 00:01:05
  To: All                                               27-Sep-99 21:33:15
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: donm@ftel.net (Don Morse)

In message <as1zbmkuxm.fsf@sci.fi> - Anssi Saari <as@sci.fi> writes:
:>
:>Lee Pearson <warp@ktn.net> writes:
:>
:>> During startup, while a dialog is displayed with the 
:>> title "Gathering Information. Please, Wait!", the following
:>> error pops up.
:>> 
:>> Critical Error
:>> TList error exception (EListError)
:>> occured at $0001048C !
:>> Terminate program?
:>> <_Yes> <_No>
:>> 
:>> Yes results in:
:>>  
:>> Exception occured:
:>> List error exception (EListError)
:>> occured at $00038BF5
:>> Program is terminated.
:>> <OK>
:>> 
:>> No results in the program exiting silently(no other errors.)
:>
:>I get the exact same thing. Warp 4, FP11 here.
:>
:>-- 
:>Anssi Saari - as@sci.fi
:>

I got that error when I installed the s/w and didn't reboot to load the
driver.
after that...  everything ok...

********************************************************
  If a million monkeys on typewriters can eventually
       type out the Bible, given enough time.
     Then Bill Gates had 25 monkeys and a week! 
********************************************************
  dmorse@pacificnet.net using Merlin and EmTec News
    ICQ 245937, AOL IM merlinof2  www.blackpalace.com
********************************************************

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Franklin interNet http://www.franklin.net (1:109/42)

+----------------------------------------------------------------------------+

From: warp@ktn.net                                      27-Sep-99 22:17:02
  To: All                                               28-Sep-99 05:15:29
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: Lee Pearson <warp@ktn.net>

On Mon, 27 Sep 1999 16:27:22 +0200 (CDT), Andreas Linde wrote:

>hello
>
>>When I try to use it, I get the 
>>following errors:
>>
>>1) "cannot open ini-file at $0001048C ! 
>>Terminate Program ?"
>>

>same error here....
>

I entered theseus on the hobbes.nmsu.edu search and found that a new version
of 
SysInfo/2 had been uploaded: sysinfo0031.zip

The description directly referenced the fixing of this error.  SysInfo/2 now 
appears to run well here.

--Lee

---
============================
Lee Pearson <warp@ktn.net>
============================


--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Wolfe Internet Access, L.L.C. (1:109/42)

+----------------------------------------------------------------------------+

From: windowssucks@hotmail.com                          28-Sep-99 14:08:28
  To: All                                               28-Sep-99 14:36:26
Subj: Re: SysInfo/2 Project - Gamma 3 (Need beta testers)

From: "The MicroSoft Down Da Drain Corp." <windowssucks@hotmail.com>

On Sat, 25 Sep 1999 12:19:35 -0400 (EDT), Bob Stan wrote:

>On Sat, 25 Sep 1999 01:08:10 +0100 (CET), Windos Sucks wrote:
>>- My Matrox G200 AGP is not correctly identified, i.e. SysInfo reports 1024
>>kB videomem instead of 8192 kB.
>
>My Matrox G200 is identified correctly with the 8192 kb of memory it has. 
>The first page leaves the drive info blank, although I have 2 WD 13 gig
>drives.  The disk tabs show the right info tho'

I'm using a G200 Marvel. Perhaps it's slightly different?

>Great utility program !

Absolutely!



--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: The MicroSoft Down Da Drain Corp. (1:109/42)

+----------------------------------------------------------------------------+

From: askbill*AT*ibm.net                                28-Sep-99 23:55:26
  To: All                                               28-Sep-99 16:15:08
Subj: Re: HPFS386, The Reality...

From: askbill*AT*ibm.net

In <37EFD7F0.71CF1240@ca.ibm.com>, on 09/27/99 
   at 04:47 PM, Brad BARCLAY <bbarclay@ca.ibm.com> said:

>askbill*AT*ibm.net wrote:
>> If that were all, it would have been/still would be cheap and cheerful
>> enough to modify "ordinary" HPFS to use a larger cache......

>	Except that the IFS model in many versions of OS/2 still uses some 16
>bit addressing.  It's an addressing limit that places a 2Mb max cache
>size HPFS.IFS, not something in the IFS itself.

By "modify" I include recompiling for 32-bit.  Relatively trivial if
nothing else is changing, (which, of course it may or may not).

>	OS/2 WARP Server for e-business introduced a new 32 bit IFS model which
>is used by JFS for using bigger cache sizes.  Theoretically, for WSEB
>users, a new HPFS driver could be coded which can take advantage of the
>larger possible cache sizes, but those in these groups using anything
>prior to WSEB wouldn't be able to use it (and those of us with WSEB can
>always use JFS for anything that benifits from big caches).
> 
I may have a skewed view, as I had WSA SMP with HPFS386, so trialed the
Aurora beta *ONLY* with HPFFS386 - since I was allowed to do so!

OTOH - JFS and LVM did not (and do not) work on my systems.  Seem to
conflict with the ASUS DA-2100 RAID virtualization.  No contest which one
was more valuable, so no JFS.

>> IMNSHO, the re-coding to run HPFS386 in Ring 0 is actually more important,
>> partly because it also applies when either a small-memory model reduces
>> the ability to take advantage of larger cache, or a high-end RAID
>> controller provides its own hardware caching.

>	This only becomes important in servers which are fielding quite a large
>number of IO requests from fast media.  On your typical client machine
>you'll hit the drive and physical controllers IO limits before you'll hit
>your CPU's limits in handling the IO requests.

For those running single-drives, no doubt.  For those running multiple
SCSI drives on multiple controllers there is a very real benefit.

The combination of many independent heads, multiple SCSI, smaller drives
(LESS swept area under the head can be an advantage) makes for a very
effective system at low cost.  

> 
>> In Weendoze & EIDE, for sure.  In OS/2 it might be more accurate to say:
>> 
>> "...is spent by the CPU doing other useful things while waiting for a
>> response from the hard drive."

>	I was assuming a one application situation where the app waits for the
>completion of IO before going on to whatever it needs to do next.  This
>was done to simplify the argument, ubt you're right - OS/2 will happily
>do other instructions for other threads while any one (or more) threads
>are blocked on IO.
> 
>> >       For those who really want to improve their disk speed, there is no
magic
>> >pill.  Go SCSI and buy a nice, quick UltraWide SCSI drive (on my Quantum
>> >Atlas 4 I'm doing between 16 and 24 Mb/s, with a 6.9ms access time :).
>> 
>> Two "magic pills" are to use multiple smaller drives (cheap but cheerful),
>> then hardware duplex (level one RAID - not so cheap).

>	For client based systems I'd say this is unnecessary.  Many modern SCSI
>drives have independent heads which can help take care of this for you
>without requiring any added hardware, while remaining entirely
>transparant to the OS itself.	
> 

Multiple heads, yes.  Multiple head positioners?  Name one, please.

>> Reads, OTOH, benefit from the ability to find identical data under EITHER
>> of two head-assemblies/rotating media sectors, one of which will
>> statistically be closer to the data than the other.

>	Well, that depends.  Sectors are only read in sequential order if you're
>using a FCFS (first come, first serve) buffer on your sector requests. 
>Most modern drive and filesystem controllers don't use FCFS on sector
>requests.  More commonly, sector requests are placed in a priority queue
>which ranks the next sector to be read based on one of a variety of
>possible criteria.  A step up is to use SSTF (shotest seek time first),
>which enqueues incoming sector requests based upon which ones take less
>effort to reach based on the current drive head positioning.  A step up
>from that is the SCAN enqueuing mechanism, which is similar to SSTF, but
>is unidirectional (ie: not just the closest sector, but the closest
>sector in the direction the head is currently moving).  There are
>variations on thi as well, incluing c-SCAN, n-step SCAN, VSCAN and CVSCAN
>(and others!).  For a more detailed discusion of these, see
>http://yaztromo.idirect.com/seekopt.html

>	The point being "statistically closer" may not matter - indeed, it can
>cause the paths to potentially lengthen.  The real benifit in the
>mechanism you propose is that you can do simultanious IO - while one
>drive is reading the first sector, the second drive can be reading the
>second.  In this sort of situation, you wind up losing about half your
>head latency on average.
> 

Close on.  Started doing it with paired CDC Wren 4's on Novell 3.xx,
Adaptec EISA 1744's or some such.  Been a while, but it has *always* paid
big dividends in reliability as well as speed.

>> None of this requires HPFS386.  All of it benefits from it, especially in
>> 256 MB or so of RAM and on dual-CPU's.

>	Well, in the end faster interfaces and faster drives give you the real
>speed increases.  Plus, anything which reduces CPU load (such as SCSI
>interfaces) is also going to give you a boost.  

And "all of the above" is best, of course.  But a handful of last-year's
close-out drives at bargain prices keep me grinning.  Four IBM Ultrastar
4.3 GB UW's arranged as two 4.3's (duplexed) on  ASUS controllers with 32
MB buffers give me better performance than a single 9.6, reduce the need
for backup, and will get upgraded to multiple 9.6 when the 18.8 pushes the
price down, etc.  Not a Luddite approach, just staying "state of the
$mart" instead of state of the art.  The controllers - limited by PCI bus
speed - last years longer than the MB & CPU.

>> Any computer which has not anticipated your next move is too slow!

>	Overall, I think we're in agreement:  speed is good :).

"Older whisky, faster horses, younger women, more money"

Bill Hacker
-- 
-----------------------------------------------------------
askbill@ibm.net (William B. Hacker, III)

Titanic '12   NYSE '29   Windows '95 and subsequent.....
-----------------------------------------------------------

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Global Network Services - Remote Access Mail & Ne
(1:109/42)

+----------------------------------------------------------------------------+

From: cfrank@rumms.uni-mannheim.de                      30-Sep-99 22:31:27
  To: All                                               30-Sep-99 21:28:08
Subj: Re: enhanced USB driver: looking for testers

From: cfrank@rumms.uni-mannheim.de (Carsten Frank)

On Wed, 29 Sep 1999 21:05:11, "Robert Lalla" 
<rlalla@stepnet.REMOVETHIS.de> wrote:

> I just completed some enhancements to the USB host controller (UHCI) driver
for OS/2.
> The following changes are included:
>  * fix VIA chipsets not working 
>  * fix screen message at boot-up
>  * support for up to 8 root hub ports (not yet verified)
>  * make transfer list access more dma-safe (cannot be verified)
> 
> I'm looking for testers that
>  * have a VIA or ETEQ mainboard chipset (Intel also welcome), or a VIA USB
addon card
>     -or-
>  * have more than 2 USB root-hub ports
>     -or-
>  * experienced previously random crashes under heavy USB load
> 
> If someone wants to test the new driver please contact me by mail:    
rlalla@stepnet.de
> 
> OHCI chipsets from ALI, OPTI, SIS or AMD are still not supported.
> 
> 
> --
> Robert Lalla, Loerrach, Germany
> 
When will be the OHCI part done, or will it never be done

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Origin Line 1 Goes Here (1:109/42)

+----------------------------------------------------------------------------+

From: mcbrides@erols.com                                30-Sep-99 18:09:20
  To: All                                               30-Sep-99 21:28:08
Subj: Re: IBM wants us to pay for java and netscape????

From: mcbrides@erols.com (Jerry McBride)

In article <SKfw30zmCGmZ-pn2-cdhddqtJK7UM@localhost>,
doug.bissett"at"ibm.net (Doug Bissett) wrote:
>On Wed, 29 Sep 1999 17:47:17, Peter Jespersen <flywheel@image.dk>
>wrote:
>
>> Business costumers will now have to pay for upgrades and bug
>> fixes...with 10-12 mio licenses this is, in short turn, good
>> business.
>> But it is bad service and therefore in the long run it is bad
>> business!
>>
>
>I agree. Way back, in about 1982, I tried to tell IBM management that
>they had better start to pay some serious attention to the IBM PC
>market (I was a service specialist, large systems, at the time). I was
>told, that "nobody would ever use a PC in a serious way. It is just a
>toy". Well, after they lost a LOT of money, for a couple of years,
>they smartened up and started getting serious abut the PC. Now, they
>are taking very much the same approach to OS/2, although I doubt if
>some of the BIG customers will let them off the hook so easily <g>.
>
>My advice: SELL your IBM shares, the price will be going down as the
>user base dries up (again). After the share price craters, buy IBM
>shares, and the new, fully supported, OS/2.
>
>Funny how history repeats itself, because nobody pays any attention to
>it (are you listening Lou???)...

Hey Doug,

You'll love this one... I ripped it from a mailing list somewhere... sorry...

--- quote ---


On Sept 16, the day IBM killed the Warp 5 client, the stock opened at $135.

Today, it closed at $120.

There are about 1.9 billion shares of IBM outstanding.

That means IBM lost $28.5 billion in capitalization or about 11% of its
total market value since that day.

--- end quote ---


Makes you wonder WHO is counting the beans at IBM. It sure ain't a stock
holder!

--

*******************************************************************************

*            Sometimes, the BEST things in life really ARE free...           
*
*       Get a FREE copy of NetRexx 1.150 for your next java project at:      
*
*                     http://www2.hursley.ibm.com/netrexx                    
*
*******************************************************************************


/----------------------------------------\
| From the desktop of: Jerome D. McBride |
|         mcbrides@erols.com             |
\----------------------------------------/

--

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: TEAM-NETREXX (1:109/42)

+----------------------------------------------------------------------------+

From: doug.bissett"at"ibm.net                           30-Sep-99 20:29:13
  To: All                                               01-Oct-99 02:22:09
Subj: Re: IBM wants us to pay for java and netscape????

From: doug.bissett"at"ibm.net (Doug Bissett)

On Wed, 29 Sep 1999 17:47:17, Peter Jespersen <flywheel@image.dk> 
wrote:

> Business costumers will now have to pay for upgrades and bug
> fixes...with 10-12 mio licenses this is, in short turn, good
> business.
> But it is bad service and therefore in the long run it is bad
> business!
> 

I agree. Way back, in about 1982, I tried to tell IBM management that 
they had better start to pay some serious attention to the IBM PC 
market (I was a service specialist, large systems, at the time). I was
told, that "nobody would ever use a PC in a serious way. It is just a 
toy". Well, after they lost a LOT of money, for a couple of years, 
they smartened up and started getting serious abut the PC. Now, they 
are taking very much the same approach to OS/2, although I doubt if 
some of the BIG customers will let them off the hook so easily <g>.

My advice: SELL your IBM shares, the price will be going down as the 
user base dries up (again). After the share price craters, buy IBM 
shares, and the new, fully supported, OS/2.

Funny how history repeats itself, because nobody pays any attention to
it (are you listening Lou???)...
******************************
From the PC of Doug Bissett
doug.bissett at ibm.net
The " at " must be changed to "@"
******************************

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Global Network Services - Remote Access Mail & Ne
(1:109/42)

+----------------------------------------------------------------------------+

From: cvopicka@erols.com                                30-Sep-99 21:47:19
  To: All                                               01-Oct-99 06:44:09
Subj: Re: IBM wants us to pay for java and netscape????

From: Ron Vopicka <cvopicka@erols.com>

> --- quote ---
> 
> 
> On Sept 16, the day IBM killed the Warp 5 client, the stock opened at $135.
> 
> Today, it closed at $120.
> 
> There are about 1.9 billion shares of IBM outstanding.
> 
> That means IBM lost $28.5 billion in capitalization or about 11% of its
> total market value since that day.
> 
> --- end quote ---


Somebody better check some facts.

On Sept 16, IBM probably opened at 131 15/16 (previous days close), the
range for the day Warp died was 132 7/16 to 129/ 9/16  closing at 130
(never traded at 135).  And today it closed at $121.

One might say IBM SURGED from 131 15/16 to 132 7/16 on the news of
Warp's demise!

Compare that to closings of HP without the Warp baggage

Sep 15 - 108.50   Sep 16 - 104   Sep 30 - 90.75

One must conclude that had HP dropped Warp from its product line, it
would have closed at 99.60 today instead of 90.75.

Gillette works too

From 41.88 to 42.50 down to 33.94 today... but they operate on razor
thin margins... right?  Without Warp instead of with Track III they
could have been at $39!

Depends on your point of view.

Ron

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Origin Line 1 Goes Here (1:109/42)

+----------------------------------------------------------------------------+

From: jeffos2@mindspring.com                            01-Oct-99 02:11:22
  To: All                                               01-Oct-99 06:44:10
Subj: Re: IBM wants us to pay for java and netscape????

From: jeffos2@mindspring.com (Jeffery Swagger)

In <SKfw30zmCGmZ-pn2-cdhddqtJK7UM@localhost>, doug.bissett"at"ibm.net (Doug
Bissett) writes:
>
>I agree. Way back, in about 1982, I tried to tell IBM management that
>they had better start to pay some serious attention to the IBM PC
>market (I was a service specialist, large systems, at the time). I was
>told, that "nobody would ever use a PC in a serious way. It is just a
>toy". Well, after they lost a LOT of money, for a couple of years,
>they smartened up and started getting serious abut the PC. Now, they
>are taking very much the same approach to OS/2, although I doubt if
>some of the BIG customers will let them off the hook so easily <g>.
>
>My advice: SELL your IBM shares, the price will be going down as the
>user base dries up (again). After the share price craters, buy IBM
>shares, and the new, fully supported, OS/2.
>
>Funny how history repeats itself, because nobody pays any attention to
>it (are you listening Lou???)...

Oh horse hockey.  OS/2's effect on IBM's bottom line is almost
non-existent.  S/390, AS/400, chip manufacturing, services and software
for the Microsoft platform dominant their revenue streams.  S/390 and
AS/400 are both essentially IBM only markets.  Either one is the
platform of choice for the explosive e-business market.  And if you
doubt that then simply compare Charles Schultz (S/390) experience vs
eBay (Sun).  IBM consulting, custom software and out-sourcing services
are high-profit and in high-demand.  IBM is the world's low-cost,
high-quality, chip manufacturer.  And guess who is the world's largest
software publisher for the Microsoft platform?  Hint:  It's not
Microsoft.  And, like or not, that is a huge market.  Of course we
haven't even touched upon the Linux potential which IBM is totally
committed to.  And who owns more intellectual property patents than
anybody else?  That right, IBM.

The bad old days of John Akers are long gone.  Whatever I believe to be
unfortunate and mis-guided decisions regarding OS/2, Gerstner has placed
IBM to be the leading technology company of the next century.  IBM
shares are definitely buy and hold.

----
Jeff

   I still miss my ex, but with this laser sight...


--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: None (1:109/42)

+----------------------------------------------------------------------------+

From: bozsi@freespace.net                               01-Oct-99 15:58:00
  To: All                                               01-Oct-99 20:01:20
Subj: Re: IBM wants us to pay for java and netscape????

From: bozsi@freespace.net (Joe Kovacs)

In <c1.2c.2SNJbP$02M@localhost.mindspring.com>, jeffos2@mindspring.com
(Jeffery Swagger) writes:
>In <SKfw30zmCGmZ-pn2-cdhddqtJK7UM@localhost>, doug.bissett"at"ibm.net (Doug
Bissett) writes:

Response in comp.os.os2.advocacy.    :-)


Joe Kovacs
Guelph Ontario Canada


--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: Water Utilities Hydraulic Analysis (1:109/42)

+----------------------------------------------------------------------------+

From: void@nospam.demon.nl                              01-Oct-99 19:20:22
  To: All                                               01-Oct-99 23:42:03
Subj: Re: IBM wants us to pay for java and netscape????

From: huug <void@nospam.demon.nl>

>>>>> "John" == John Thompson <nospam@savebandwidth.invalid> writes:

 John> But at the same time, IBM has announced support for open source
 John> initiatives:

That's GOOD news.  Now all they have to do is add Warp Client to the
zone :)

-- 
Note: I reserve the right to publish or return unsolicited, harassing or 
annoying (e-)mail.  /hy:x/                               PGP keyID: 0xDF28F4C1
"All are strange but thee and me; but sometimes thee acts funny."- Chuck Blake

--- WtrGate+ v0.93.p7 sn 165
 * Origin: Usenet: voluntary (1:109/42)

+----------------------------------------------------------------------------+

+============================================================================+
