What Linux needs to topple M$

What Linux needs to topple M$

Post by Michael Pop » Sun, 16 Mar 1997 04:00:00



Allrightie, this has been a long time in coming.  This is just
IMHO and reflects nobody's oppinions but my own.  OK, here we go:

1) OLE-like system:
It's true.  Think about it - unless you know a LOT about the typesetting
languages (and maybe not even then) you can't display a picture you drew
from, say, XPaint, or a spreadsheet, or anything else.  Theoretically
it wouldn't be that hard - every type of application would have a
'viewer' applet - one that drew to an allocated GC on demand when
supplied
a filename and/or data.  The viewer applets would be free, but you
wouldn't be able to edit the data without the application.  I wouldn't
be supprised if this is how Microsoft works OLE, or at least it's
similar.

2) A good IDE:
Don't you sometimes get tired of:
nedit
xterm:
        make
        errors:
        #(*&$
nedit
joe Makefile
#*($
nedit
make
nedit
make
#*($
nedit
make
#(*#
gdb
#(*$($*(#*$(#*$(*$#(*#!?!
Ctrl-Alt-Delete

Well, I can say quite assuredly that I do.  One of the few things about
Windows that I envy for Linux (aside from OLE - though M$'s OLE is a
bloated
bag of #(*$, surely we can do better, we've done everything better than
M$ up to this point... :) ) is a good, solid IDE.  One with project
view,
multiple source windows, compiling, etc. from a standard Motif or
Motif-like interface.  Compiling, running, etc.  Oh yeah, NO RC FILES!
This should have a dialog configuration!  And MDI, too...

3) "Hiding" the multi-user aspects
OK, this SEEMS far-fetched first, but hear me out.
I've actually tried this before:
edit /etc/passwd
make an account (user) and make sure the password field is empty
go to the end of the rc.* chain
insert:
login user
startx

This jumps you right into X.  The user can ignore the cryptic start-up
messages
like always with DOS.  Utilities run with shell scripts could be
suid'ed/or temporarily log in as root and allow you to control
everything.
Of course, in the setup the inclusion of said login account, login
manuvers,
and utilities is COMPLETELY optional (say, maybe for a 'desktop system'
configuration).

4) Standardized interface
Although not enforced (like Micro$oft) it should be encouraged to have
a Windows-like interface, reguardless of the toolkit.  This will make
it easier on just about everyone...  

Wait, wait, I've got a BETTER idea!!!
INSTEAD of hardcoding the location of, say, the menu bars, or the scroll
bars, we could have a toolkit read them in from a configuration file!
That
way everything's exactally where everyone wants it for themselves!!!

5) Clipboard
We need one for everything (though the Motif clipboard works OK), that's
compatible with the Linux OLE I proposed a moment ago.  'nuff said.

6) Control Panels
Although we have the good Glint (only really good part of Red Hat IMHO)
we need a control-panel for /etc and other
key configuration repositories.  Also, there should be a text-only
version for criticle stuff (i.e., you were logged in as root, were
drinking late one night, don't remember what you did but you woke up
at the console and /usr/X11R6 had ceased to exist...) and text-only
stuff (i.e. Bash configuration).

7) No disk caching
This should be a valid option (and should automatically be enabled in
the
'desktop configuration' option of installation that is afforementioned)
that
would automatically sync the disk during read/writes, thus preventing
unsightly data destruction.  Most users will no doubt thing that
you can just turn off a Linux system and think everything is
"hunky-dorey".
In fact, I can attest to this - my sister almost whacked the reset
button
when I was in Linux so she could reboot into Windows.  Stopped her just
in time...

8) Automated floppy mounting/CD-ROM mounting
Although I've been told we have this via the supermount package, I don't
think it works quite right with the current kernels.  This should
probably
be maintained by someone in the kernel team (no, I'm not volunteering,
because
if I were to do so it would destroy half the test-users machines when
they
boot up... I just sort of have that 'magic programming charm' :)

9) Shared library aliasing
There should be a way to 'alias' a SO as another.  For example, Motif
2.0
is compatible with Motif 1.2, yet you need libXm.so.1.2.4 to run Motif
1.2.4 dynamically linked applications, even if you have libXm.so.2.0.x !
Although I've heard they have their reasons for making it not work like
this you should still be able to 'alias' it (but cover the thing in
'danger, not liable for blah blah blah' messages :) )

10) Office suite -
Not the bloated stuff that's coming out for Windows.  I'm pretty sure
that we can muster together a decent Word Processor with most core
functions
in 1 executable (>5 MB) if dynamically linked.  Same for a spreadsheet
and
Access/DBase-like database application.  Throw in a few extras (i.e.
spellchecker, OLE-ish stuff for Linux, paint program) we'd probably
be able to put something together in the 40-50 meg range, installed.

11) Graphics built-in to the kernel/Real-Time stuff
Although this is being worked on, I believe that this is quite
neccessary.  This is more a pat-on-the-back to the developers of these
kernel add-ons than anything else.

12) In-application configuration
It isn't very fun for any novice (and many hackers) to hunt through
RC files to change word-wrap and the like.  A good configuration system
built into the application to re-write RC files as neccessary would be
invaluable.

13) Drag-n-Drop
We have this, but nobody (and I mean NOBODY) supports the same
protocols!!!
OffiX has one, Motif has another, and there's gotta be a dozen more...
everyone
seems to write their own whenever they make a program!  Can't we agree
on something as fundamental as this?...

OK, this seems like a lot but I think it's worth it.  Before you go and
say "He's full of crap, he's not even doing any of this stuff, just
bossing us around" I'd like to say that's not true.  I am in the process
of trying to learn X/Motif programming so that I can (attempt to) write
the afforementioned IDE.  I'm not a terribly good programmer, but I'm
trying,
at least.

Also, please be aware that this is my OPPINION.  I'm not trying to have
a "Holier than thou" stance or a bossy stance.  At the most I'm trying
to
give the programmers out there something to write, stuff which I feel
Linux sorely needs if Linux will become a viable desktop OS for the
masses.  Yes, the masses.  I believe the world will actually be a better
place if we can bring Linux to the common man yet retain the hacking
qualities
of Linux, which is mutually benificial (hackers can use their skills for
programming jobs(let's face it, UNIX programmers aren't as in-demand as
Windows programmers in most places), common people get a good solid
desktop interface, etc.)

I await any response you can give me, but please, no flames are
neccessary.
Of course, this being an advocacy group... :)
Also, if you e-mail me please don't send the whole message as a quote
unless absolutely neccessary.  I'm running low on disk space :P

-Michael Pope

---------------------------------
|!PUBLIC NOTICE ! PUBLIC NOTICE!|
| This is copyright 1997 by     |
| Michael Pope.  Rights are     |
| given to be distributed freely|
| in any medium to all but the  |
| Microsoft corporation.  They  |
| may purchase a license for    |
| $25,000 per instance for the  |
| same rights, except THEY must |
| quote the article in full     |
| original form before          |
| modifications, commentary,    |
| etc. and subsequent           |
| redistribution.  This license |
| purchase applies to Microsoft |
| corporation, subsidiaries,    |
| Microsoft Press, Microsoft    |
| Network, SCO Inc. and any     |
| other company in which        |
| Microsoft founded and/or has  |
| more than .75% stock in said  |
| corporation.                  |
| All rights reserved.          |
---------------------------------

 
 
 

What Linux needs to topple M$

Post by Keith E. Moo » Mon, 17 Mar 1997 04:00:00


In article <332B8A44.BCD...@inreach.com>, Michael Pope wrote:
>Allrightie, this has been a long time in coming.  This is just
>IMHO and reflects nobody's oppinions but my own.  OK, here we go:

>1) OLE-like system:
>It's true.  Think about it - unless you know a LOT about the typesetting
>languages (and maybe not even then) you can't display a picture you drew
>from, say, XPaint, or a spreadsheet, or anything else.  Theoretically
>it wouldn't be that hard - every type of application would have a
>'viewer' applet - one that drew to an allocated GC on demand when
>supplied
>a filename and/or data.  The viewer applets would be free, but you
>wouldn't be able to edit the data without the application.  I wouldn't
>be supprised if this is how Microsoft works OLE, or at least it's
>similar.

  Solaris, DEC-Unix and Linux will be the first three *IX systems
to get Active-X.  (Believe it or not, SCO will come _later_).

>2) A good IDE:
>Don't you sometimes get tired of:

[snip]

 Actually, I find Visual Slickedit for Linux to be a nice system.
also, you should really use DDD or xxgdb, 100x better than just gdb.

>Well, I can say quite assuredly that I do.  One of the few things about
>Windows that I envy for Linux (aside from OLE - though M$'s OLE is a
>bloated
>bag of #(*$, surely we can do better, we've done everything better than
>M$ up to this point... :) ) is a good, solid IDE.  One with project
>view,
>multiple source windows, compiling, etc. from a standard Motif or
>Motif-like interface.  Compiling, running, etc.  Oh yeah, NO RC FILES!
>This should have a dialog configuration!  And MDI, too...

  Visual Slick has this.  The only thing it does not have is Makefile
generation, but anyone who has been programming for any time has
a skeleton Makefile laying around that they just fill in.

>3) "Hiding" the multi-user aspects
>OK, this SEEMS far-fetched first, but hear me out.
>I've actually tried this before:
>edit /etc/passwd
>make an account (user) and make sure the password field is empty
>go to the end of the rc.* chain
>insert:
>login user
>startx

>This jumps you right into X.  The user can ignore the cryptic start-up
>messages

Uhm... you will probably get nasty flames from this, but I'll try
to be a little level-headed:

  Look at the docs for XDM.  (Generally just switching to a new
runlevel will set it up).  Under RedHat, you can use the UserManager, and
it has the option of a Blank Password.  You can use the RunLevel Editor
to switch to X-Windows Only also.

All OS's today, including Windows'95 (if there is a network connection)
ask for a Userid and password, so it is really pointless anyway to try
to get rid of this.

>like always with DOS.  Utilities run with shell scripts could be
>suid'ed/or temporarily log in as root and allow you to control
>everything.
>Of course, in the setup the inclusion of said login account, login
>manuvers,
>and utilities is COMPLETELY optional (say, maybe for a 'desktop system'
>configuration).

 Please... let's not revert to DOS.  We are not competing with insecure
OS's, even Microsoft has had to start tightening it's security on
even it's desktop OS's.  Corporate America does not trust it's
employees.

>4) Standardized interface
>Although not enforced (like Micro$oft) it should be encouraged to have
>a Windows-like interface, reguardless of the toolkit.  This will make
>it easier on just about everyone...  

 It's called the CDE, and it's available for Linux.

>Wait, wait, I've got a BETTER idea!!!
>INSTEAD of hardcoding the location of, say, the menu bars, or the scroll
>bars, we could have a toolkit read them in from a configuration file!
>That
>way everything's exactally where everyone wants it for themselves!!!

  Even TheNextLevel does this.

>5) Clipboard
>We need one for everything (though the Motif clipboard works OK), that's
>compatible with the Linux OLE I proposed a moment ago.  'nuff said.

   That's a whole different discussion.

>6) Control Panels
>Although we have the good Glint (only really good part of Red Hat IMHO)
>we need a control-panel for /etc and other
>key configuration repositories.  Also, there should be a text-only
>version for criticle stuff (i.e., you were logged in as root, were
>drinking late one night, don't remember what you did but you woke up
>at the console and /usr/X11R6 had ceased to exist...) and text-only
>stuff (i.e. Bash configuration).

  Guess you don't use RedHat.

>7) No disk caching
>This should be a valid option (and should automatically be enabled in
>the
>'desktop configuration' option of installation that is afforementioned)
>that
>would automatically sync the disk during read/writes, thus preventing
>unsightly data destruction.  Most users will no doubt thing that
>you can just turn off a Linux system and think everything is
>"hunky-dorey".
>In fact, I can attest to this - my sister almost whacked the reset
>button
>when I was in Linux so she could reboot into Windows.  Stopped her just
>in time...

   Uhm... you're just going off into lala land now... actually you
COULD set the sync deamon to 1 second, but it would cut some of the
incredible speed improvements Linux has over other OS's.

>8) Automated floppy mounting/CD-ROM mounting
>Although I've been told we have this via the supermount package, I don't
>think it works quite right with the current kernels.  This should
>probably
>be maintained by someone in the kernel team (no, I'm not volunteering,
>because
>if I were to do so it would destroy half the test-users machines when
>they
>boot up... I just sort of have that 'magic programming charm' :)

  Why? Just use the MTOOLS and you don't need to mount the floppies.

As for the CDROM, I have had NT lose the CDROM (i.e. is always
unavailable until next reboot).

I prefer having to mount the CDROM, however I DO have a button on
my desktop to do it, and of course I allow normal users to mount the
CDROM (nosuid, of course).

>9) Shared library aliasing
>There should be a way to 'alias' a SO as another.  For example, Motif
>2.0
>is compatible with Motif 1.2, yet you need libXm.so.1.2.4 to run Motif
>1.2.4 dynamically linked applications, even if you have libXm.so.2.0.x !
>Although I've heard they have their reasons for making it not work like
>this you should still be able to 'alias' it (but cover the thing in
>'danger, not liable for blah blah blah' messages :) )

  Wow, you really have no clue.  Just toss in a symbolic link, actually
if the Motif you got didn't sym-link to 1.2, you got a bad distribution.

>10) Office suite -
>Not the bloated stuff that's coming out for Windows.  I'm pretty sure
>that we can muster together a decent Word Processor with most core
>functions
>in 1 executable (>5 MB) if dynamically linked.  Same for a spreadsheet
>and
>Access/DBase-like database application.  Throw in a few extras (i.e.
>spellchecker, OLE-ish stuff for Linux, paint program) we'd probably
>be able to put something together in the 40-50 meg range, installed.

 Applix, StarOffice, Clique, CorelJava, others still coming.

>11) Graphics built-in to the kernel/Real-Time stuff
>Although this is being worked on, I believe that this is quite
>neccessary.  This is more a pat-on-the-back to the developers of these
>kernel add-ons than anything else.

  That would be the single largest mistake ever made.  Even Gates
has admitted that moving the Graphics to the kernel may have been
a mistake with NT, it has caused him nothing but grief since it's
release.

>12) In-application configuration
>It isn't very fun for any novice (and many hackers) to hunt through
>RC files to change word-wrap and the like.  A good configuration system
>built into the application to re-write RC files as neccessary would be
>invaluable.

  Up to the program.

>13) Drag-n-Drop
>We have this, but nobody (and I mean NOBODY) supports the same
>protocols!!!
>OffiX has one, Motif has another, and there's gotta be a dozen more...
>everyone
>seems to write their own whenever they make a program!  Can't we agree
>on something as fundamental as this?...

  Fundamental to you, "fluff" to others.

- Show quoted text -

>OK, this seems like a lot but I think it's worth it.  Before you go and
>say "He's full of crap, he's not even doing any of this stuff, just
>bossing us around" I'd like to say that's not true.  I am in the process
>of trying to learn X/Motif programming so that I can (attempt to) write
>the afforementioned IDE.  I'm not a terribly good programmer, but I'm
>trying,
>at least.

>Also, please be aware that this is my OPPINION.  I'm not trying to have
>a "Holier than thou" stance or a bossy stance.  At the most I'm trying
>to
>give the programmers out there something to write, stuff which I feel
>Linux sorely needs if Linux will become a viable desktop OS for the
>masses.  Yes, the masses.  I believe the world will actually be a better
>place if we can bring Linux to the common man yet retain the hacking
>qualities

   I could care less if it becomes viable for the Masses, I would like
it to be viable for alot of people, but Linux is a superb OS, that
has more going for it than any other OS out there.  Pigeon-holing it
into a single type of Interface because it would be easier for someone
who doesn't know how to set his/her wall-clock would be the
wrong thing to do.

>of Linux, which is mutually benificial (hackers can use their skills for
>programming jobs(let's face it, UNIX programmers aren't as in-demand as
>Windows programmers in most places), common people get a good solid
>desktop interface, etc.)

>I await any response you can give me, but please, no flames are
>neccessary.
>Of course, this being an advocacy group... :)
>Also, if you e-mail me please don't send the whole message as a quote
>unless absolutely neccessary.  I'm running low on disk space :P

>-Michael Pope
>d...@inreach.com

  Please, research the possibilities before posting, and cut down
your line width.

--
-- Keith Moore
   President
   KMA Computer Solutions, Inc.
...

read more »

 
 
 

What Linux needs to topple M$

Post by John Steve » Mon, 17 Mar 1997 04:00:00



Quote:

>1) OLE-like system:

Why bother?  Supposedly, MS is writing this for us. . .

And if Linux ends up supporting GNUStep, this will be done better using
GNUSteps distributed objects.

Me, I want CORBA 2.0 as well.

Quote:>2) A good IDE:
>Don't you sometimes get tired of:
>nedit
>xterm:
>    make
>    errors:
>    #(*&$

No.  The cycle isn't any different with an IDE than it is on a command
line.

If you can put up with it, try Emacs.  It'll give you a type of IDE.

Quote:>3) "Hiding" the multi-user aspects
>I've actually tried this before:
>edit /etc/passwd
>make an account (user) and make sure the password field is empty
>go to the end of the rc.* chain
>insert:
>login user
>startx

Don't access the Internet while doing this.  Odds are that you didn't
also turn off your machines ability to accept telnet/ftp connections,
so just anybody who wanted to could waltz right in. . .

Quote:>Utilities run with shell scripts could be
>suid'ed/or temporarily log in as root and allow you to control
>everything.

!!!!!

Yup.  Rip out your modem, any networking cards and make sure that
nobody who isn't a trained Unix admin has access to this box!

In combo with the above idea of a non-passworded account, you've
just allowed the Internet interloper full access to root. . .

An easier way of doing this is simply running everything as root.

Quote:>4) Standardized interface

This can already be done.  Simply buy all Motif programs.

Quote:>5) Clipboard
>We need one for everything (though the Motif clipboard works OK), that's
>compatible with the Linux OLE I proposed a moment ago.  'nuff said.

IE, we need one that is compatible with the new MS OLE?  Probably
that will become an issue.

Quote:>7) No disk caching

Yeek!  You do know what this would do to performance, don't you?

Quote:>Most users will no doubt thing that
>you can just turn off a Linux system and think everything is
>"hunky-dorey".

When I install a Linux box, I disable reset and install a locking power
button.  Simple solution.  I then install a login that does a
shutdown -h now.  This solves most problems.

Quote:>In fact, I can attest to this - my sister almost whacked the reset
>button
>when I was in Linux so she could reboot into Windows.  Stopped her just
>in time...

Sounds like you need to do the same thing as I do.

Quote:>8) Automated floppy mounting/CD-ROM mounting

Mounted as what?  Where? With what kind of privs?  Or, are you suggesting
that this fit the 'single user' paradigm again?

Quote:>9) Shared library aliasing

If you are feeling adventurous, ignoring version numbers is a pretty
good way to crash a program.  I wouldn't add this as a facility,
as this kind of thing should remain very hard to do.

Quote:>10) Office suite -

Try Applix.  $200 from RedHat, right now.

Quote:>13) Drag-n-Drop
>We have this, but nobody (and I mean NOBODY) supports the same
>protocols!!!

All Motif programs do.  Buy pure Motif and stop worrying about it.

Quote:>OffiX has one, Motif has another, and there's gotta be a dozen more...
>everyone
>seems to write their own whenever they make a program!  Can't we agree
>on something as fundamental as this?...

Probably not.  One of the strengths of Linux is that it makes a great
platform for research, testing, trying out new ideas.  If you want
just one DnD protocol, stick with just one.

Quote:>give the programmers out there something to write, stuff which I feel
>Linux sorely needs if Linux will become a viable desktop OS for the
>masses.  Yes, the masses.

The masses don't want Linux.  Note that what you have suggested is making
Linux into Windows.  Why not simply stick with Windows?  MS does the kind
of thing you want done much better than Linux will every do it.

Quote:>I believe the world will actually be a better
>place if we can bring Linux to the common man yet retain the hacking
>qualities of Linux, which is mutually benificial

The reason we have so many different Window Managers, So many new
GUI development teams, so many different DnD protocols, so many
programs what use text files for configuration is that Linux is, indeed,
an OS the allows hackers free reign, and in fact, encourages them.

What you want, and what Linux is, are two radically different things.

Quote:>(hackers can use their skills for
>programming jobs(let's face it, UNIX programmers aren't as in-demand as
>Windows programmers in most places), common people get a good solid
>desktop interface, etc.)

Unix programmers are in high enough demand that I have never had any
problem finding a job.  And the common people already have a good,
solid desktop.  Use Windows, or Mac.

John S.

 
 
 

What Linux needs to topple M$

Post by Tim Smi » Mon, 17 Mar 1997 04:00:00



Quote:>2) A good IDE:
>Don't you sometimes get tired of:
>nedit
>xterm:
>    make
>    errors:
>    #(*&$
>nedit
>joe Makefile
>#*($
>nedit
>make
>nedit
>make
>#*($
>nedit
>make
>#(*#
>gdb
>#(*$($*(#*$(#*$(*$#(*#!?!
>Ctrl-Alt-Delete

That's how I work on Windows! (except it's wmake or nmake, wdw, and vi or viw).

Quote:>4) Standardized interface
>Although not enforced (like Micro$oft) it should be encouraged to have
>a Windows-like interface, reguardless of the toolkit.  This will make
>it easier on just about everyone...  

Microsoft doesn't enforce the interface (neither does Apple on the Mac): they
just make it easier, by providing code that does much of the work for you if
you use it.  What enforces the interface on both Windows and the Mac is the
market.  If someone finds they have to read the manual or help file to find
out how to create a new document, for example, they will not think highly of
your product.

Quote:>7) No disk caching
>This should be a valid option (and should automatically be enabled in
>the
>'desktop configuration' option of installation that is afforementioned)
>that
>would automatically sync the disk during read/writes, thus preventing
>unsightly data destruction.  Most users will no doubt thing that
>you can just turn off a Linux system and think everything is
>"hunky-dorey".
>In fact, I can attest to this - my sister almost whacked the reset
>button
>when I was in Linux so she could reboot into Windows.  Stopped her just
>in time...

Did she realize it was Linux and not DOS?  If you weren't in X at the time,
she may have thought you had left DOS running.  Windows users have pretty
much all figured out by now that Windows caches and it is not a good idea
to go and hit the reset button, so I don't see that it is a major problem
when they come to Linux.  Just tell them that Linux caches like Windows
does, so be sure to run the shutdown.

Better would be to change Linux so that ctrl-alt-del asks the user if the
user wishes to shut down or restart.  (Right now, on most Linux's, it just
initiates shutdown).  Better would be to enhance ctrl-alt-del to be an
actual trusted path that asks if you want to shut down, restart, or login
on the console (killing any processes currently logged in on the console).
I believe a trusted path is needed for C2 security, so this would be useful
to add anyway.

But I digress.  Anyway, killing performance by getting rid of caching in
the default setup is not wise.  Better would be to have a logging filesystem
that survives arbitrary power loss.

Quote:>10) Office suite -
>Not the bloated stuff that's coming out for Windows.  I'm pretty sure
>that we can muster together a decent Word Processor with most core
>functions
>in 1 executable (>5 MB) if dynamically linked.  Same for a spreadsheet
>and
>Access/DBase-like database application.  Throw in a few extras (i.e.
>spellchecker, OLE-ish stuff for Linux, paint program) we'd probably
>be able to put something together in the 40-50 meg range, installed.

Actually, that's near the range of Office if you get rid of all the
extras.  The extras add a lot of offices' "bloat".  E.g., the online
documentation for each component is large (many megabytes), and they
have optional clip art collections, and stuff like that.

--Tim Smith

 
 
 

What Linux needs to topple M$

Post by Tim Smi » Mon, 17 Mar 1997 04:00:00



>Don't access the Internet while doing this.  Odds are that you didn't
>also turn off your machines ability to accept telnet/ftp connections,
>so just anybody who wanted to could waltz right in. . .

For a desktop machine whose only connection to the outside world is via a
dialup PPP or SLIP link, I don't see why telnet/ftp serving are enabled in
the default configuration.

--Tim Smith

 
 
 

What Linux needs to topple M$

Post by Tim Smi » Mon, 17 Mar 1997 04:00:00



Quote:>All OS's today, including Windows'95 (if there is a network connection)
>ask for a Userid and password, so it is really pointless anyway to try
>to get rid of this.

On nice feature in Windows 95 is that if you establish a dial up internet
connection, and you have any directories or volumes (or printers?) on your
computer shared on your local Microsoft network, you get a warning that
strangers out on the net may have access, and it offers to temporarily
disable sharing until the dialup connection goes away.  (Of course, it would
be nicer if it were smart enough to disable sharing over the dialup connection
while keeping it on the LAN).

This could be a useful addition to Linux (it goes without saying, of course,
that in Linux, it would be done right, only cutting off services over the
dialup, leaving the LAN alone).

Quote:>  That would be the single largest mistake ever made.  Even Gates
>has admitted that moving the Graphics to the kernel may have been
>a mistake with NT, it has caused him nothing but grief since it's
>release.

It's a philosophical mistake, but what grief has it caused?  I know that
there was a bug in the parameter checking of some calls, but that's the
only problem I've heard of, and it has helped performance quite a bit.

--Tim Smith

 
 
 

What Linux needs to topple M$

Post by Arun Sharm » Mon, 17 Mar 1997 04:00:00



        >> 2) A good IDE: Don't you sometimes get tired of: nedit
        >> xterm: make errors: (*&$

        John> No.  The cycle isn't any different with an IDE than it
        John> is on a command line.

        John> If you can put up with it, try Emacs.  It'll give you a
        John> type of IDE.

How about a GUI built around Emacs (or for that matter any editor,
including, yes, vi) to provide the IDE ? Automating make file
generation and having it tightly coupled with the de* would be
major pluses.

        -Arun
--

 
 
 

What Linux needs to topple M$

Post by Nathan Han » Tue, 18 Mar 1997 04:00:00




> >4) Standardized interface
> >Although not enforced (like Micro$oft) it should be encouraged to have
> >a Windows-like interface, reguardless of the toolkit.  This will make
> >it easier on just about everyone...

Windows-like interface?!? Heck no. UNIX consortiums have tried for
years to get a standardised interface. There's another one in the
works: the KDE project http://www.veryComputer.com/. Hopefully this one will
have more luck than the others.

Quote:> >7) No disk caching

Scream, heck no. Imagine the performance loss! It'd be hideous.

Quote:> Better would be to change Linux so that ctrl-alt-del asks the user if the
> user wishes to shut down or restart.  (Right now, on most Linux's, it just
> initiates shutdown).  Better would be to enhance ctrl-alt-del to be an
> actual trusted path that asks if you want to shut down, restart, or login
> on the console (killing any processes currently logged in on the console).
> I believe a trusted path is needed for C2 security, so this would be useful
> to add anyway.

You can do this yourself by editting /etc/inittab and changing the
program called by "ctrlaltdel" from shutdown to whatever. Though I
assume you already knew this, perhaps other readers don't.

Quote:> But I digress.  Anyway, killing performance by getting rid of caching in
> the default setup is not wise.  Better would be to have a logging filesystem
> that survives arbitrary power loss.

Yes, a much better solution. Check out the Linux Logging Filesystem
home page http://www.veryComputer.com/~c-cook/prof/lfs/.

Quote:> >10) Office suite -
> >Not the bloated stuff that's coming out for Windows.  I'm pretty sure
> >that we can muster together a decent Word Processor with most core
> >functions
> >in 1 executable (>5 MB) if dynamically linked.  Same for a spreadsheet
> >and
> >Access/DBase-like database application.  Throw in a few extras (i.e.
> >spellchecker, OLE-ish stuff for Linux, paint program) we'd probably
> >be able to put something together in the 40-50 meg range, installed.

Office packages abound for linux. Staroffice, WordPerfect, Applixware
are off the top of my head. The Linux software pages have much more
comprehensive lists of commercial products. I prefer lyx myself which
is free (GPL-free) and very complete. I believe StarOffice is free
for non-commercial use but you might like to make sure before bugging
the developers or accidentally ripping them off.

For painting you can't go past the GIMP.

Quote:> Actually, that's near the range of Office if you get rid of all the
> extras.  The extras add a lot of offices' "bloat".  E.g., the online
> documentation for each component is large (many megabytes), and they
> have optional clip art collections, and stuff like that.

Actually lyx on linux isn't small, but it isn't "bloated". You add up
lyx and*and tex and xdvi and dvips and ghostview and ghostscript
and pretty soon you hit the 50MB margin without doing anything fancy.
Of course, I would have all those other packages installed anyway, so
the extra "burden" of lyx is almost nothing.

--
Open mind for a different view, and nothing else matters.

 
 
 

What Linux needs to topple M$

Post by Douglas Ridgw » Tue, 18 Mar 1997 04:00:00


Just a few comments...

: 6) Control Panels
: Although we have the good Glint (only really good part of Red Hat IMHO)
: we need a control-panel for /etc and other
: key configuration repositories.  

The problem is not that there are no sysadmin tools, it's that there
are too many, with no one tool dominating. Red Hat has its various
control panel tools, Caldera uses LST, there's Jaques Gelinas' linuxconf,
etc.

: 12) In-application configuration
: It isn't very fun for any novice (and many hackers) to hunt through
: RC files to change word-wrap and the like.  

App configuration is sort of up to the app. Some have fairly nice systems,
eg texconfig for tetex, and so on. But also look at The Dotfile Generator,
a general purpose system for .rc configuration file editing.

: 7) No disk caching
: This should be a valid option  [...]

man mount. You want -o sync. BSD fans prefer this to be the default,
Linux fans disagree for performance reasons. Personally, I've never been
burned by hitting the Big Red Button anyway. Just takes longer starting up.

: 8) Automated floppy mounting/CD-ROM mounting

Agree. This is one of the things I miss about Nextstep, just dropping
something in the drive and having it appear on the desktop. I also know
about supermount, but since it's not default in Red Hat, I've never bothered.
It's easier to type 'mount whatever' than download, compile, install and
configure a new package.

: 11) Graphics built-in to the kernel/Real-Time stuff

Soft realtime you can do now, hard real time requires RT-Linux. I don't
understand the arguments for and against doing graphics in kernelspace,
but enough people believe it to be the Right Thing that it's being implemented
(the GGI project).

Now, getting OpenGL built into XFree would be something really nice.

--
Douglas Ridgway
Department of Physics and Institute for Nonlinear Science
University of California, San Diego
http://inls.ucsd.edu/~ridgway/

 
 
 

What Linux needs to topple M$

Post by Matthew Borows » Tue, 18 Mar 1997 04:00:00




Quote:>Allrightie, this has been a long time in coming.  This is just
>IMHO and reflects nobody's oppinions but my own.  OK, here we go:

Yeah, but why use Motif as the standard interface? It's commercial
software and we should try to stay away from this type of thing, if
Win95 users are to get into Linux without paying too much.

------------------------------------------------------------
* Matthew Borowski, http://mkb.home.ml.org/                *
------------------------------------------------------------
* "Macintoshes have to be smart computers --               *
* they must make up for their users' lack of intelligence."*
------------------------------------------------------------

 
 
 

What Linux needs to topple M$

Post by Mats Andtback » Tue, 18 Mar 1997 04:00:00



Quote:>1) OLE-like system:

AIEE, don't you know what that does to security?! as well leave root
without a password and telnetd enabled.

CORBA, now, that might be something. especially if somebody figures
out how to properly fit Java into that picture.

Quote:>2) A good IDE:

emacs. vi and GNU screen. make.
what more could anybody want?

Quote:>3) "Hiding" the multi-user aspects

so make just the one account for yourself. one username to remember,
one password apart from root's, no problem.

Quote:>4) Standardized interface

you do know what a *y pig CDE is, don't you?
ugly, too. anything remotely connected to anything remotely connected
to Motif seems to be, for some reason.

Quote:>5) Clipboard
>We need one for everything (though the Motif clipboard works OK), that's
>compatible with the Linux OLE I proposed a moment ago.  'nuff said.

all you need do is solve the perpetual X11 drag-n-drop problem (one of
the jillion biggest, worstest flaws about X) and that'll be a snap.
good luck. better off just trying to mercy kill X, though.

Quote:>6) Control Panels

i like afterstep's control panel for configuring afterstep.
button that pops up vi on ~/.steprc, that is.

Quote:>Although we have the good Glint (only really good part of Red Hat
>IMHO) we need a control-panel for /etc and other key configuration
>repositories.

no we don't. wanna know why? 'cos you'll never make it smart enough to
deal with human-readable configuration files decently, is why. even
the red hat tools down this alley have their worst flaws precisely
there. ever seen the printcaps they generate? eeeeevil...

now, misunderstand me right, human-readable configuration files are a
Good Thing. they let you fix a broken machine with just a text editor
and some common sense. it just doesn't make sense to have a
complicated piece of software carefully generate such a file to (1) be
syntactically right, (2) do what you wanted it to, and (3) preferrably
still make sense to a human reader, just so that a nearly as
complicated parser can read it in and make binary configuration data
from it when whatever-the-file-controls is run. you're trying to
defeat yourself doing that, you're bound to lose.

Quote:>Also, there should be a text-only version for criticle stuff

already is. vi.

Quote:>7) No disk caching

and with the performance you'd be getting, you'll suggest we all s*
disk drives for C-cassette tapes next, right? or paper tape?

actually, what you _want_ here is to mount all your filesystems sync.
which is a perfectly valid thing to do, in certain really weird
situations. but this one isn't one.

Quote:>that would automatically sync the disk during read/writes, thus
>preventing unsightly data destruction.  Most users will no doubt
>thing that you can just turn off a Linux system and think everything
>is "hunky-dorey".

you're trying to make the system fool-proof. that's bound to fail,
because it's the wrong thing to do. what you should do, is make the
system fool-*free*.

Quote:>8) Automated floppy mounting/CD-ROM mounting

what _i_ want, is floppy drives without mechanical eject buttons, so
no idjit can pop out a mounted floppy. other than that, look into amd.

Quote:>9) Shared library aliasing

ever heard of ln(1)?

Quote:>10) Office suite -
>Not the bloated stuff that's coming out for Windows.  I'm pretty sure
>that we can muster together a decent Word Processor with most core
>functions in 1 executable (>5 MB) if dynamically linked.

there's something wrong with that size reading. you used ">" where you
wanted "<", and "5" where should be "1". HTH.

hint hint: the first halfway decent wp programs i learned to use, fit
on floppies. 360KB floppies.

Quote:>11) Graphics built-in to the kernel

are an abomination.

Quote:>/Real-Time stuff

is nice, but i'm unconvinced RT doesn't really deserve its own OS
entirely. sure, if somebody wants to put in the work and give it a
spin, let them, i'm just doubtful if the end result will be worth it.

Quote:>12) In-application configuration

if *YOU* wanna rewrite every program ever written...

[...]

Quote:>Also, if you e-mail me please don't send the whole message as a quote
>unless absolutely neccessary.  I'm running low on disk space :P

if you often use 25-line .sig's, no *y wonder, either.
--
        "...it's all wrong
         but it's alright..."          -- Clapton
 
 
 

What Linux needs to topple M$

Post by Markus Vuo » Tue, 18 Mar 1997 04:00:00



>Yeah, but why use Motif as the standard interface? It's commercial
>software and we should try to stay away from this type of thing, if
>Win95 users are to get into Linux without paying too much.

Right!

--
*************************************************************************
* MARKUS VUORI, tekn.yo, Lappeenranta University Of Technology, FINLAND *
*                  IRC: MaVu    http://www.lut.fi/~mvuori               *
*************************************************************************

 
 
 

What Linux needs to topple M$

Post by david parso » Tue, 18 Mar 1997 04:00:00






>>Allrightie, this has been a long time in coming.  This is just
>>IMHO and reflects nobody's oppinions but my own.  OK, here we go:

>Yeah, but why use Motif as the standard interface?

   Because it's the closest thing to a standard that X Windows has?

                 ____

                  \/

 
 
 

What Linux needs to topple M$

Post by david parso » Tue, 18 Mar 1997 04:00:00



Quote:>3) "Hiding" the multi-user aspects
>OK, this SEEMS far-fetched first, but hear me out.
>I've actually tried this before:
>edit /etc/passwd
>make an account (user) and make sure the password field is empty
>go to the end of the rc.* chain
>insert:
>login user
>startx

  I don't think this is the best way of doing things.  I'd instead have
  X start up automatically and then write a program that would become
  the default user, then run their .xinitrc.

  I don't know how popular this would be, because even Windows makes
  you log in before you can access shared resources.

Quote:>7) No disk caching
>This should be a valid option (and should automatically be enabled in the
>'desktop configuration' option of installation that is afforementioned) that

   Not even Windows 95 gives you this.

   It would be better to help with one of the projects that's attempting
   to port a log structured or ordered-update filesystem to Linux, and
   then you could have a system where you can pull the plug without
   unsightly fscks when you reboot.

                 ____

                  \/

 
 
 

What Linux needs to topple M$

Post by Richard Kettlewel » Wed, 19 Mar 1997 04:00:00



>7) No disk caching

RTFM mount(8) - in particular, `-o sync'.

But experience suggests it's completely unnecessary for many uses:
we've had a number of Linux machines suffering regular power cuts and
not lost a byte yet (OK so I once lost INN's history file, but that
was easily rebuilt from the spool).

--
Richard Kettlewell               http://www.elmail.co.uk/~richard/

 
 
 

1. CNN: Is Linux poised to topple Microsoft?

Headline News
Is Linux poised to topple Microsoft?
September 3, 2002 Posted: 1:59 PM EDT (1759 GMT)
http://www.cnn.com/2002/TECH/09/03/hln.wired.linux/index.html
+By Renay San Miguel
+CNN Headline News
+
+(CNN) -- During the Cold War, the initials ABM used to mean
+Anti-Ballistic Missile. In the late '90s, they stood for Anybody
+But Microsoft, a reaction to the fact that Bill Gates' Windows
+operating system was in 90 percent of the world's computers and
+critics didn't like the restrictions Microsoft Corp. placed on
+computer companies that licensed its software.
+
+But now Microsoft is a convicted monopolist, forced to ease up on
+those restrictions. The biggest beneficiaries of the New
+Millennium ABM Club may be proponents of Linux, the open-source
+operating system, long considered to be as potentially disruptive
+to Microsoft's dominance as a missile strike on Communist-era
+Moscow.

Well, I suppose that you could describe Microsoft's Palladium
so called "Digital Rights Management" efforts ...
"From XP on the PC to CE on the PDA,an iron curtain has descended
 across the Content..."

David Mohring - "There is the solution which I respectfully offer
                to you in this Address to which I have given the
                title "The System called Linux"".

2. problems with DOSEMU0.60.1 and X

3. Can Linux topple the Microsoft empire?

4. NEC Superscript printer HELP

5. CNN: Is Linux poised to topple M$ ?

6. apache: how can I map all uri requests on domain to one cgi script?

7. SCO's 64 bit Gemini with Merced will topple WinNT

8. Linux 2.2.1 : bogus high traffic with snmp/mrtg

9. NEED HELP in figuring what's needed for LINUX.

10. Help needed - Win98 --> Linux needs IPC$ Password ?

11. m$-win guy needs linux help-need content fitering software

12. Linux Needed/mailing lists needed

13. Need on Linux Boot Needed