Linux-Advocacy Digest #566, Volume #26 Wed, 17 May 00 21:13:06 EDT
Contents:
Re: progamming models, unix vs Windows (R.E.Ballard ( Rex Ballard ))
Re: Top 10 Reasons to use Linux ("Jim Ross")
Re: Desktop use, office apps ("Jim Ross")
Re: But.... didn't they say it wasn't Outlook's fault? (Alan Boyd)
Re: Desktop use, office apps ("Jim Ross")
Re: Why do I need Linux? (Christopher Browne)
Re: Desktop use, office apps ("Jim Ross")
Re: Desktop use, office apps ("Jim Ross")
Re: Desktop use, office apps (JEDIDIAH)
Re: Why do I need Linux? (JEDIDIAH)
----------------------------------------------------------------------------
From: R.E.Ballard ( Rex Ballard ) <[EMAIL PROTECTED]>
Subject: Re: progamming models, unix vs Windows
Date: Wed, 17 May 2000 23:42:47 GMT
In article <[EMAIL PROTECTED]>
> [EMAIL PROTECTED] (Pete Goodwin) wrote:
> [EMAIL PROTECTED] (R.E.Ballard ( Rex Ballard )) wrote in
<8fucab$92l$1
> @nnrp1.deja.com>:
>
> >Actually, part of the problem is that Microsoft
> >didn't get multitasking right until Windows 2000,
> >and even then it's a kludge between threads and
> >processes which has been nicknamed "fabrics".
>
> Do you consider NT didn't get multitasking right?
> I'm not disagreeing with you,
> I'm just curious on what you base that statement.
Compared to UNIX or Linux, the overhead for multitasking
on NT is quite substantial. The cost of creating a process,
initializing the process, context switching, and cache management
are all much higher on NT than they are for Linux or UNIX.
Much of this goes back to the fact that the core of NT's kernel
is based on VMS. VMS had the same issues relative to UNIX.
To run OpenVMS effeciently, the administrator would try to
create a huge pool of preinitialized processes and then farm them
to the initiated applications.
Microsoft was able to hide the problems by placing the emphasis
on threads, encouraging programmers to spawn hundreds or thousands
of threads under a single application. This created a little memory
management problem as race conditions and deadlocks could either lead
to application failures that dropped hundreds of connections
concurrently, or deadlocks that would cause a process serving hundreds
of users to freeze and deny all connections.
Furthermore, the application, or application libraries, had to
manage the memory and communication across hundreds of theads.
This certainly made programming and application development more
difficult. It also made "locking up the exe" much more desirable
to reduce the risk of a new plug-in taking out the system.
Windows 2000 with MTS is definitely a step in the right direction.
Under Windows 2000, programmers are strongly encouraged to use
apartment threading, and to allow MTS to manage the memory. MSMQ
can be used for safe and effective interprocess communication. In
terms of functionality, there's very little difference between MSMQ
and UNIX IPC, which is based on message queues. Even pipes and stdio
are implementations which use message puts to fill the buffer, and
use message gets to pull messages from the buffer and start the
receiving application. The advantage of data streams over
prescribed length messages is that you can use smaller standard
size messages and parse the smaller stream.
Windows 2000 has also had an overhaul of it's DLL layouts. The
most frequently used methods have been moved closer together,
and the less frequently used exception and help messages have
been moved out of the main cachable area. This reduces thrashing
of the L1 and L2 caches. It's still not as efficient as Linux and
UNIX (which were written in c and therfore not directly associated
with objects), but it has improved performance by about 20%.
Windows NT also had some rather intense garbage collection duties.
Each JVM had it's own garbage collector, and each application had
it's own garbage collector - largely due to the expense of kernel
based memory allocation. Windows 2000 fabrics still use object pools,
but it's a nice comprimize to compensate for the inefficient malloc
of NT/2000.
Windows 2000 context switching is more efficient as well. More of
the kernel and core methods are left in cache, and more of the core
library functions are left in the L1 cache. Linux is especially
crafty in it's tweaking of L1 and L2 cache storage. It's one of
the advantages it has over more traditional versions of UNIX.
> >Microsoft considers it's "components" to
> >be library calls (COM objects) that can be
> >Linked together by corporations (Primarily
> >Microsoft) to create huge monolithic
> >applications. For load balancing,
> >these calls can be fed to remote systems
> >by calling DCOM instead of COM. Essentially,
> >your calling a DLL which makes RPC calls
> >to the remote library. The application
> >programmer is barely aware of the change,
> >but the administrators become very aware.
>
> Microsoft are not the only ones building
> monolithic applications; the last
> place I worked had a 10 MByte executable,
A really good example of this type of bloatware
is Netscape Navigator. Because it uses static linked
Mosaic, core calls, object oriented widgets organized
by object type (instead of by frequency of execution),
and has poor garbage collection, the initial load,
with a blank page displayed, is about 16 megabytes. This
can rapidly grow to 32 megabytes or more as memory is
consumed but not garbage collected for Motif gadgets.
The new Navigator 6.0, which was based on open source and supported
by standard open source libraries optimized for execution starts with
an initial load image of about 6 megabytes and shrinks to about a 3
megabyte executable area (rest in swap space) and a modest buffer
area.
> and that was just the core.
> Everything was COM, and there was a DCOM component in the works.
It's likely that they will want to migrate directly to COM+ and
avoid the hassles of DCOM memory and thread management.
> >The fundamental distinction of Microsoft
> >platforms is that applications must perform
> >a huge portion of the "Operating System"
> >functions, either through library calls
> >(static, DLL, or DCOM) or through application
> >functions. MSMQ is a half-step in the right
> >direction, but falls far short of System V
> >Interprocess communications, local-domain
> >sockets, and pipelines, including named pipes.
> >Even simple sockets through localhost tend
> >to increase the overhead substantially.
>
> I don't know enough to compare Microsoft
> Message Queue (I assume that's what you mean by MSMQ)
> but doesn't NT have IPC, sockets, pipes and named
> pipes?
The IPC is extremely limited, and imposes substantial
overhead on NT. Part of the problem is that the kernel
must perform a number of context switches for each relay.
In a traditional bucket brigade datastream chain, this
can mean several dozen context switches to parse a relatively
simple and small datastream.
More often, Microsoft prefers the "object model" in which the
instances are fully initialized and self contained. In other words,
if I want do edit a word document containing a bitmap object,
the entire bitmap would remain in memory for the duration of the
edit session. This can be a real problem when you have a help
object that contains numerous bitmaps or mpeg videos. Full size
buttons in 32bit color are pretty to look at, but when an application
has 300 of these buttons, the overhead can be quite substantial.
> >Also, the availability of a number of scripting
> >languages and incremental compilers or configured
> >interpreters can provide very high performance
> >with ease of customization.
>
> With perhaps just one snag - the scripts
> can be a bit difficult to read.
True. Many scripts are deliberately obfuscated. There
was once a contest for the best "Obfuscated C" and the
ideas were often used to confuse would-be hackers intent
on cracking code intended only for configuration support.
> >Because there are 500 million Microsoft based Personal
> >Computers and Microsoft still provides minimal backward
> >compatibility. The application that ran on Windows 3.1
> >has been recompiled and enhanced to run on NT, but still
> >opens "C:\a\b\c.d". The installation scripts still require
> >specific drives, and the few applications that are "well behaved"
> >enough to run under the NT version of MS-DOS emulator still
> >require drive names.
>
> And so many applications dump so
> much junk in \Windows\System it's
> unbelievable. There was a recent article
> in Computing about the DLL Hell -
DLL Hell is probably one of the best examples of flaws
in the Microsoft development paradigm. Every year, Microsoft
redefines "well behaved" applications. This was necessary
because many programs would break if hardware or software
were upgraded. The most classic example was copy-protection
software that used software timing loops based on a 4.77 mhz
clock on an 8088 processor to twiddle with the disk drive registers
to create copy protection sectors. When the PC/XT came out
with an 8088 all the copy protection software had to be recalibrated.
Of course, this also meant that legal owners of legal software
were having their software removed or disabled because of an
assumption.
Ironcially Microsoft eventually figured out that changing DLLs
was a good business practice. It forced OEMs to use the latest
software, it forced third party vendors to either be very nice
to Microsoft or it forced competitors out of the market, and it
forced end users to upgrade software and hardware every 2-3 years,
often discarding perfectly functional computers that just wouldn't
run the new software.
Originally, you could safely bet that you software would still be
useful until you purchased the next version of the software.
Unfortunately, Microsoft discovered that service packs were a way
to impose "upgrade" using "free" upgrades. In fact, Service Pack 4
was supposed to be a mandatory upgrade, required to be Y2K compliant,
and was released with only a few weeks before the "big day". Many
third party applications were disabled by the service pack, especially
Microsoft competitor products like Lotus Notes, Netscape Navigator,
Eudora, some versions of CORBA, some versions of Java, and some
Office Suites like WordPerfect Office, Lotus SmartSuite, and
most applications developed using the Borland/Inprise compilers
and toolkits. Many of these torpedos were removed in Service pack 5,
and many of the third party vendors offered free service packs or
upgrades of their own. In some cases, the patches even torpedoed
Microsoft Office.
> version control of those DLL's that
> make up the system, and which so many
> applications written in Visual C++ depend on.
Both Visual C++ and Visual Basic have suffered horribly from
version upgrades. Even worse hit were the applications developed
using non-microsoft tools and IDEs. This ultimately resulted
in initiatives by many third-party vendors to support Linux.
At least if you're torbedoed by Linux, you can get the source
code and find out what you need to change in your application.
Backward compatibility is very important to the Linux community.
It's still possible to run pre-ELF applications on modern Linux
machines.
> >Windows 95/fat 32 still doesn't support real long-names. The
> >implementation of long-names is based on an alias file which is
> >used to translate long-names to 8.3 filenames. I think NTFS has
> >real long-names.
>
> It's transparent enough... mostly,
> until you try to do something before
> Windows starts then you need the "short" name!
Numerous backup and restore utilities get confused, many
shareware programs get hopelessly lost, and many MS-DOS
command line interfaces can actually corrupt the names.
Also, when burning CD-ROMs, you may have to rename long names
to shorter names, and this may break other links.
> >Linux combines the best features of UNIX (streams, processes, fast
> >context switching, and X11) with some of the best features of
Windows
> >(friendly GUI interfaces for applications and administration, helps,
> > wizards, and hints). Linux also agreessively went after the
> >configurability of the PC in terms of hardware and software. There
> >are some peripherals that aren't supported, but more and more
> >OEMs and After-market vendors are discovering that Linux support
> >sells hardware. Even if users don't use Linux full-time, they
> >are insisting on the ability to run Linux at least some of the time.
>
> I must have missed something here - Linux still seemed to have
> configuration spread all over the place.
Generally, there are a few places you need to back up. The
/root directory, the /etc directory, the /usr/etc directory,
and the /usr/local/etc directory. It's also a good idea to
back-up the home directory and the /usr/lib/X11/app-defaults
directory. This can generally be managed using simple shell
scripts and shar scripts.
Some applications, especially those application centric
applications which are ports of Windows applications, may have
unique configuration files, but the files needed to restore
network configuration, services, and other standard system features
are included in a handful of directories that would fit on a
single floppy.
> Pete
>
--
Rex Ballard - Open Source Advocate, Internet
I/T Architect, MIS Director
http://www.open4success.com
Linux - 60 million satisfied users worldwide
and growing at over 1%/week!
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: "Jim Ross" <[EMAIL PROTECTED]>
Subject: Re: Top 10 Reasons to use Linux
Date: Wed, 17 May 2000 19:59:42 -0400
Noname <[EMAIL PROTECTED]> wrote in message
news:MozU4.4471$[EMAIL PROTECTED]...
>
> 12. You grew an allergy to Microsoft products. You want
> to try something non-microsoft to see if that can stop
> your symptoms of itchy watery eyes and runny nose
Or just get some work done.
>
> 11. You want to steal something without getting caught
> by police. Linux gives you the lawful protection to
> such adventure
You can't steal what's free. I guess you missed that important piece of
information.
Jim
------------------------------
From: "Jim Ross" <[EMAIL PROTECTED]>
Subject: Re: Desktop use, office apps
Date: Wed, 17 May 2000 20:05:26 -0400
Ray <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> On Tue, 16 May 2000 15:53:32 GMT, R. Christopher Harshman
> <[EMAIL PROTECTED]> wrote:
> >
> >The problem: We cannot find office productivity software that fits our
> >needs, and I was wondering why I had not heard more about the glaring
> >problems that exist with the current offerings. In a nutshell:
> >
> >StarOffice
> >Far too slow to load. We're using just the applications (launching
> >`soffice staroffice.private:starwriter` for instance, to use just the
> >word-processor without the desktop). Even on the fastest of our
> >workstations, a Celeron 466 with a brand-new UDMA/66 hard drive, it
> >takes almost a minute to load. Once loaded, it's more or less
> >responsive enough to use, but the users we've had test the configuration
> >have universally complained about the wait.
>
> If SO 5.1 is really taking that long to load then something's really
wrong.
> It takes between 10 & 15 seconds here on an AMD k6-2-300 with 64MB ram.
> This machine serves up SO and a few other apps. to the other machines here
> (running remotely via X). Additional instances of SO only take around
> 5 seconds or so to load. Heck, even my laptop (a P120 with 48MB) loads SO
> in under 30s.
>
> --
> Ray
Should a 10 to 15 seconds be considered a normal loadtime?
Word 97 wouldn't take that long and it is a better package than SO. 2
seconds max to load Word.
Staroffice does still feel sluggish on my new AMD 850 128MB, ATA/66.
It's obviously the software that's slow, not the hardware.
Jim
------------------------------
From: Alan Boyd <[EMAIL PROTECTED]>
Subject: Re: But.... didn't they say it wasn't Outlook's fault?
Date: Wed, 17 May 2000 19:05:11 -0500
mlw wrote:
>
> http://www.theregister.co.uk/000516-000003.html
>
> So, only after two really really bad viruses, Microsoft is going to
> "innovate" by adding features that are so common sense, that a U.S.
> congressman, not even known for technical savvy, suggested in a hearing.
>
> Duh!
>
> Break up Microsoft, now!!!!
>
> P.S. I would use the word "ubiquitous" as opposed to "popular" when
> describing "Lookout." Popular has a common meaning of being well liked.
Here's part of a story that I found interesting.
http://news.bbc.co.uk/hi/english/sci/tech/newsid_750000/750456.stm
Microsoft was keen to point out that it was
only limiting the functions within Outlook, not
closing a security hole.
"Given the global impact of the ILOVEYOU virus
and the growing threat of malicious hackers,
we strongly believe we must take the
unprecedented step of limiting certain popular
functionality in Outlook to provide a significant,
additional security option for our customers,"
said Steven Sinofsky, senior vice president of
Microsoft Office.
They're going to provide an additional security option. In other words
it didn't exist before. And it's a "significant" option, so I guess
it's an important one. But this option doesn't plug a security hole, it
just limits functionality.
George Orwell must work for MS 'cause this sure sounds like NewSpeak to
me.
(I hope it was NewSpeak, that's how I remember it.)
--
"I don't believe in anti-anything. A man has to have a
program; you have to be *for* something, otherwise you
will never get anywhere." -- Harry S Truman
------------------------------
From: "Jim Ross" <[EMAIL PROTECTED]>
Subject: Re: Desktop use, office apps
Date: Wed, 17 May 2000 20:09:35 -0400
Tim Koklas <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Now, that's the first time I hear someone claim that Word is crap. But
> again ... equal alternatives?
Exactly.
Word runs very fast, opens most any document type, and has tons of features,
and fonts look greeat
It's very very hard to compete with.
Word 97 is as nearly a perfect wordprocesor in my opinion.
I wouldn't change a thing.
AbiWord seems to be going in the right direction, but is missing many things
and fonts always seems bad in X. I'm sick of that at this point.
Maybe in 3 years X with get anti-aliased and look normal for once.
Oh and all Motif apps should go away fast. They looked sick before and sick
now.
Jim
>
> [EMAIL PROTECTED] wrote:
>
> > Word is an abomination. They should have quit in '94.
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Why do I need Linux?
Reply-To: [EMAIL PROTECTED]
Date: Thu, 18 May 2000 00:13:25 GMT
Centuries ago, Nostradamus foresaw a time when [EMAIL PROTECTED]
would say:
>In article <M11U4.86394$[EMAIL PROTECTED]>,
> [EMAIL PROTECTED] wrote:
>
>> Furthermore, almost all the computer systems you likely encounter are,
>> under the covers, becoming "more like UNIX" over time.
>
>It appears that since Unix is perceived by non-technical people as
>being a stable system, so most vendors use this as a marketing point
>and want to put as many Unix features into their systems as possible
>(as if adding fork() takes out all of the bugs in your OS).
To be sure, Unix _is_ monotonically more stable than any of the
"desktop platforms." They're not seeing an effect that is outright
incorrect.
>In many systems this has dramatically hindered technical progress of
>the OS. For example, until about 10 years ago or so, VMS could handle
>dates through the year 9999. Since the introduction of Unix-like
>concepts (particularly TCP/IP and DECwindows, which were ported from
>Unix and written in C), VMS is now only certified through 2038 (or
>2106, actually).
That is indeed rather unfortunate.
>Windows (which is already a primordial soup of other OS'es influence)
>and MacOS (which is also about to be Unixified) will likely also have
>disastrous results as the become more and more indistiguishable from
>Unix.
On the other hand, it is not nearly so evident that _this_ represents
an outright "disaster."
"UNIXifying" Windows doesn't seem to me to be _that_ much of a
disadvantage; it provides the possibility that Microsoft might be able
to grab some code from *BSD or, as it fits, from Linux-related
systems, and actually make Windows both more powerful and more robust.
The same seems true for MacOS. Having a Real OS Kernel underneath, as
opposed to the former CMT "mess," looks to be an improvement.
Would it have been realistic to expect either company to adopt VMS?
Microsoft _did_ bring in a bunch of the Prism developers, but it
_looks_ like it's a fresh team working on W65K these days.
>Not only is there a personal argument against all OS'es being the same
>(such that there is nothing more to learn, and no more creative choices
>to be made), but there is a very serious security argument against
>going to one OS. Despite having different sources, many different Unix
>systems are plagued by the same bugs because they have similar design
>decisions. For example, in 1994 there was a bug which affected both AIX
>and Linux systems, whereby logging in as the user "-froot" would give
>root access to any user. Although AIX and Linux had no code in common,
>they had similar implementations due to similar requirements, and both
>feel victim to the same bug. The more widespread Unix is, the more
>vulernable world security is.
I agree about 90% with this.
The last decade has been a _disaster_ from an OS research point of
view. Microsoft was "buying out" anybody with viable alternatives to
just about anything but Windows, with the result that there has been a
dearth of useful research on "what might come next."
It was doubtless a good decade for Rich Raschid; he doubtless got some
decent stock options out of becoming VP of Research at Microsoft.
Hopefully there will be some interesting things come from the ashes;
as Microsoft has been under legal pressure, this has prevented
remaining _overly_ rapacious. University of Utah seems to have a
reasonably active OS research group (FluxOS, OSKit, Fluke), as does
University of Dresden (L4, Fiasco). The UPenn work on EROS also
sounds promising.
The unfortunate thing is the [virtual] deaths of:
a) BSD research;
b) Mach research at CMU;
c) OS research at MIT (exokernel last "stable" release published in
1998...)
which represent places that used to have _substantial_ and
_influential_ OS-related projects.
--
Do not meddle in the affairs of dragons, for you are crunchy and taste
good with ketchup.
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: "Jim Ross" <[EMAIL PROTECTED]>
Subject: Re: Desktop use, office apps
Date: Wed, 17 May 2000 20:13:39 -0400
<[EMAIL PROTECTED]> wrote in message news:8fubef$7nl$[EMAIL PROTECTED]...
> Applix 5.0 works just fine for me.
>
Filters seem to suck.
The whole thing feels clunky somehow, even though it uses GTK now.
Jim
------------------------------
From: "Jim Ross" <[EMAIL PROTECTED]>
Subject: Re: Desktop use, office apps
Date: Wed, 17 May 2000 20:21:05 -0400
R. Christopher Harshman <[EMAIL PROTECTED]> wrote in message
news:8fuu2u$uid$[EMAIL PROTECTED]...
> I suppose I should have been more clear initially. Our requirements:
> * Microsoft Office compatibility, import and export.
> * Runs and loads with a reasonable amount of performance across a wide
> variety of workstation hardware, from a 32MB 486DX2/66 to a 64MB Celeron
> 466. (The hardware Linux is supposed to increase the usefulness of.)
That's the first lie.
The Linux people don't expect you to want a GUI too.
So no Linux is slow also on an old machine like Win9X was since you are
talking about a desktop system,
and Linux people seem very focused only on server systems.
>
> My issue is predominantly that machines that performed acceptably well
> under Windows 9x with MS Office 97 either do not perform reasonably well
> , or are not reasonably MS Office compatible, when running Linux and
> currently available office solutions.
>
> Applix 5.0 is fine but not Office compatible (not really; my users won't
> hassle with .rtf files). The fact that StarOffice is quick on your
> PIII-500 with 128MB of RAM doesn't help the fact that our budget doesn't
> allow for such beasts, much as I covet them.
>
> Is there a way to work this, or should I bite the bullet and reload
> Win9x on (at least the lower end of) these workstations?
I would.
I've never considered a Linux system a good desktop replacement for
Win9X/NT.
The thing the Linux people don't understand or refuse to agree on is the
idea of 1.
Yes it is limiting, but you can streamline the hell out the system that way.
Win9X is 20MB of CABs.
Suse is what 6 CDs/1 DVD.
That is not a good formula for desktop success my friends.
One toolkit, and make that toolkit kick serious arse.
This explains 100% why Linux has a 4% market share of desktop systems.
That will not likely grow beyond that if Linux people don't heed this
advice.
>
> For those who dispute the validity of Office + Windows outperforming
> Linux + StarOffice or Linux + Corel WordPerfect Suite 2000, I'll be
> happy to upload MPEG video clips showing the relative performance of
> each, running on identical hardware. (And unlike Microsoft's trial
> tapes, mine won't be doctored. =)
>
> It's not a matter of a few seconds; minutes more like it across the
> board, especially on older hardware but noticeable on the newer gear as
> well.
>
> Yours,
> Chris
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
I would let Linux improve for a few years and try again later.
Linux as a desktop was useless back in 97/98 when I first looked at it.
It's somewhat better now.
Jim
------------------------------
From: [EMAIL PROTECTED] (JEDIDIAH)
Subject: Re: Desktop use, office apps
Date: Thu, 18 May 2000 00:23:20 GMT
On Wed, 17 May 2000 20:05:26 -0400, Jim Ross <[EMAIL PROTECTED]> wrote:
>
>Ray <[EMAIL PROTECTED]> wrote in message
>news:[EMAIL PROTECTED]...
>> On Tue, 16 May 2000 15:53:32 GMT, R. Christopher Harshman
>> <[EMAIL PROTECTED]> wrote:
>> >
>> >The problem: We cannot find office productivity software that fits our
>> >needs, and I was wondering why I had not heard more about the glaring
>> >problems that exist with the current offerings. In a nutshell:
>> >
>> >StarOffice
>> >Far too slow to load. We're using just the applications (launching
>> >`soffice staroffice.private:starwriter` for instance, to use just the
>> >word-processor without the desktop). Even on the fastest of our
>> >workstations, a Celeron 466 with a brand-new UDMA/66 hard drive, it
>> >takes almost a minute to load. Once loaded, it's more or less
>> >responsive enough to use, but the users we've had test the configuration
>> >have universally complained about the wait.
>>
>> If SO 5.1 is really taking that long to load then something's really
>wrong.
>> It takes between 10 & 15 seconds here on an AMD k6-2-300 with 64MB ram.
>> This machine serves up SO and a few other apps. to the other machines here
>> (running remotely via X). Additional instances of SO only take around
>> 5 seconds or so to load. Heck, even my laptop (a P120 with 48MB) loads SO
>> in under 30s.
>>
>> --
>> Ray
>
>Should a 10 to 15 seconds be considered a normal loadtime?
It's certainly is within reasonable tolerances for something
so superficial.
>Word 97 wouldn't take that long and it is a better package than SO. 2
>seconds max to load Word.
>
>Staroffice does still feel sluggish on my new AMD 850 128MB, ATA/66.
>It's obviously the software that's slow, not the hardware.
SO5 is fine on my K63 for operations comparable to that which
I might carry out on Office 97. In areas where it exceeds the
functionality of Office 97, it does get sluggish.
--
In what language does 'open' mean 'execute the evil contents of' |||
a document? --Les Mikesell / | \
Need sane PPP docs? Try penguin.lvcm.com.
------------------------------
From: [EMAIL PROTECTED] (JEDIDIAH)
Subject: Re: Why do I need Linux?
Date: Thu, 18 May 2000 00:25:32 GMT
On Thu, 18 May 2000 00:13:25 GMT, Christopher Browne <[EMAIL PROTECTED]> wrote:
>Centuries ago, Nostradamus foresaw a time when [EMAIL PROTECTED]
>would say:
>>In article <M11U4.86394$[EMAIL PROTECTED]>,
>> [EMAIL PROTECTED] wrote:
[deletia]
>>In many systems this has dramatically hindered technical progress of
>>the OS. For example, until about 10 years ago or so, VMS could handle
>>dates through the year 9999. Since the introduction of Unix-like
>>concepts (particularly TCP/IP and DECwindows, which were ported from
>>Unix and written in C), VMS is now only certified through 2038 (or
>>2106, actually).
>
>That is indeed rather unfortunate.
Especially since the VAX 11/80 instruction set includes
support for 64bit integers...
[deletia]
--
In what language does 'open' mean 'execute the evil contents of' |||
a document? --Les Mikesell / | \
Need sane PPP docs? Try penguin.lvcm.com.
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.advocacy) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
