Linux-Misc Digest #247, Volume #24 Sun, 23 Apr 00 15:13:02 EDT
Contents:
gcc "make error" when building a linux-x-djgpp cross compiler (AL FLOYD)
psion (Ian Mortimer)
Re: localhost & domainname change (Herb Stein)
Re: bash_profile setting (Floyd Davidson)
gcc error while building a linux-x-djgpp cross compiler (AL FLOYD)
Re: Summing Up File Sizes (Herb Stein)
Re: HOT HD (Charlie Brown)
Re: Corel Linux (Rod Smith)
templates and g++ (Donovan Rebbechi)
Re: core dumps (or the lack of them) (Leonard Evens)
please help with network ("Bigoffthewall")
Freeedom to Innovate Network (Tony Lawrence)
Re: linux programs, tools and functionality for windows? (Dave Brown)
FS: toshiba 430CDT and ZipDrive $480 and FDC settlement of $100! (Chris T)
Re: HOT HD ("Justin R. Miller")
EMERGENCY!!! (mh)
Re: How Microsoft inhibits competition & innovation (Se�n � Donnchadha)
----------------------------------------------------------------------------
From: AL FLOYD <[EMAIL PROTECTED]>
Subject: gcc "make error" when building a linux-x-djgpp cross compiler
Date: Sun, 23 Apr 2000 10:46:42 -0600
This is the log file from running "make > make.log 2>&1" under gcc build
directory.
Can anyone explain to me the error that follows:
gcc -DCROSS_COMPILE -DIN_GCC -g -O2 -DHAVE_CONFIG_H -o collect2
collect2.o tlink.o hash.o intl.o underscore.o version.o
obstack.o ../libiberty/libiberty.a
cp xgcc gcc-cross
/usr/src/djgpp/cross/gcc/gcc/xgcc -B/usr/src/djgpp/cross/gcc/gcc/
-B/usr/local/i586-pc-msdosdjgpp/bin/
-I/usr/local/i586-pc-msdosdjgpp/include -dumpspecs > tmp-specs
mv tmp-specs specs
echo "void __foo () {}" > dummy.c
/usr/src/djgpp/cross/gcc/gcc/xgcc -B/usr/src/djgpp/cross/gcc/gcc/
-B/usr/local/i586-pc-msdosdjgpp/bin/
-I/usr/local/i586-pc-msdosdjgpp/include -DCROSS_COMPILE -DIN_GCC -g
-O2 -I./include -c dummy.c
i586-pc-msdosdjgpp-ar rc libgcc1.null dummy.o
make[1]: i586-pc-msdosdjgpp-ar: Command not found
make[1]: *** [libgcc1.null] Error 127
make[1]: Leaving directory `/usr/src/djgpp/cross/gcc/gcc'
make: *** [all-gcc] Error 2
I have the file "i586-pc-msdosdjgpp-ar" sitting in my /usr/local/bin and
/usr/local/i586-pc-msdosdjgpp/bin directories. Can you tell me why it is
not finding it.
------------------------------
From: Ian Mortimer <[EMAIL PROTECTED]>
Subject: psion
Date: Sun, 23 Apr 2000 17:48:54 +0000
Hi all,
Does anyone have any applications to link a psion 3a to linux ?
I've searched freshmeat and can't find anything suitable .
Rgds,
Ian.
------------------------------
From: [EMAIL PROTECTED] (Herb Stein)
Subject: Re: localhost & domainname change
Date: Sun, 23 Apr 2000 16:51:33 GMT
For RedHat Linux, change the following files:
/etc/hosts
/etc/resolv.conf
/etc/sysconfig/network
/etc/sysconfig/network-scripts/ifcfg-eth0
If you are hosting virtual sites (probably not), also change:
/etc/sysconfig/ipaliases
In article <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
> Hi there! How do I change my hostname and domainname correctly w/o
>having trouble using pine mail. I changed the hostname is etc/hosts,
>etc/sysconfig/network, and etc/HOSTNAME, but how do I change the
>domainname? I tried to change my domainname but that gave me some
>problems, like being unable to send mail to another user in pine. Note
>that I am not in a network... Please send all the neccessary steps.
>Thanks....
>
>Linux user
>
--
Herb Stein
The Herb Stein Group
www.herbstein.com
[EMAIL PROTECTED]
314 215-3584
------------------------------
From: Floyd Davidson <[EMAIL PROTECTED]>
Subject: Re: bash_profile setting
Date: 23 Apr 2000 08:09:05 -0800
[EMAIL PROTECTED] (Dave Brown) wrote:
>In article <[EMAIL PROTECTED]>, Floyd Davidson wrote:
>>[EMAIL PROTECTED] () wrote:
>>...
>>>Yes yes yes.
>>>
>>>1. Because it works.
>>>2. Because it dosn't abbreviate the home directory to ~
>>
>>Garsh, never occured to me that anybody would *want* that. ;-)
>>
>
>The problem with \w is that it abbreviates any directory, so
>you may not know if you're in /bin or /usr/bin or /usr/X11R6/bin...
You are thinking of \W not \w. One provides the basename, the
other gives a full path but abreviates /home with ~/.
>The other problem is that you may need to switch to a korn
>shell (I occasionally do), which does not recognize
>bash-specific features. Using $PWD is transparent to both. (I
>also create a HOST variable, and use $HOST in the prompt as
>well.) If you operate in directories with long pathnames, you
>can always include a newline in the prompt.
There are better ways. Use ~/.bash_profile for bash and
~/.profile for ksh only, for example. But I prefer using a
common ~/.profile for both bash and ksh and setting ENV and
BASH_ENV to different files for ksh and bash.
Either way bash and ksh definitions of PS1 do not interfere with
each other. That frees bash from the restrictions of ksh. It
has the added advantage of allowing the addition of virtually
any shell.
Here is my ~/.profile:
if [ "${0}" = "-bash" -o `basename "${0}"` = "bash" ]
then
profile BASH_ENV=$HOME/.bashrc
profile ENV=$HOME/.bashrc
. $HOME/.bashrc
fi
if [ "${0}" = "-su" -o `basename "${0}"` = "su" ]
then
profile BASH_ENV=$HOME/.bashrc
profile ENV=$HOME/.bashrc
. $HOME/.bashrc
fi
if [ "${0}" = "-sh" -o "${0}" = "sh" ]
then
ENV=$HOME/.shrc
. $HOME/.shrc
fi
if [ "${0}" = "-ksh" -o "${0}" = "ksh" ]
then
ENV=$HOME/.kshrc
. $HOME/.kshrc
fi
--
Floyd L. Davidson [EMAIL PROTECTED]
Ukpeagvik (Barrow, Alaska)
------------------------------
From: AL FLOYD <[EMAIL PROTECTED]>
Subject: gcc error while building a linux-x-djgpp cross compiler
Date: Sun, 23 Apr 2000 11:00:27 -0600
This is the log file from running "make > make.log 2>&1" under gcc build
directory.
Can anyone explain to me the error that follows:
gcc -DCROSS_COMPILE -DIN_GCC -g -O2 -DHAVE_CONFIG_H -o collect2
collect2.o tlink.o hash.o intl.o underscore.o version.o
obstack.o ../libiberty/libiberty.a
cp xgcc gcc-cross
/usr/src/djgpp/cross/gcc/gcc/xgcc -B/usr/src/djgpp/cross/gcc/gcc/
-B/usr/local/i586-pc-msdosdjgpp/bin/
-I/usr/local/i586-pc-msdosdjgpp/include -dumpspecs > tmp-specs
mv tmp-specs specs
echo "void __foo () {}" > dummy.c
/usr/src/djgpp/cross/gcc/gcc/xgcc -B/usr/src/djgpp/cross/gcc/gcc/
-B/usr/local/i586-pc-msdosdjgpp/bin/
-I/usr/local/i586-pc-msdosdjgpp/include -DCROSS_COMPILE -DIN_GCC -g
-O2 -I./include -c dummy.c
i586-pc-msdosdjgpp-ar rc libgcc1.null dummy.o
make[1]: i586-pc-msdosdjgpp-ar: Command not found
make[1]: *** [libgcc1.null] Error 127
make[1]: Leaving directory `/usr/src/djgpp/cross/gcc/gcc'
make: *** [all-gcc] Error 2
I have the file "i586-pc-msdosdjgpp-ar" sitting in my /usr/local/bin and
/usr/local/i586-pc-msdosdjgpp/bin directories. Can you tell me why it is
not finding it.
------------------------------
From: [EMAIL PROTECTED] (Herb Stein)
Subject: Re: Summing Up File Sizes
Date: Sun, 23 Apr 2000 16:55:33 GMT
Also consider du which can go through an entire directory tree.
In article <8cvlvp$rfc$[EMAIL PROTECTED]>, "Peter T. Breuer"
<[EMAIL PROTECTED]> wrote:
>Jeff Susanj <[EMAIL PROTECTED]> wrote:
>: know how many blocks of storage or even better how many Kbytes they consume.
>: We want to do this fairly often. 'ls' will give me the information for a
>: single file but it will not provide a summary like the DOS 'dir' command
>
>just count their bytes!
>
> wc -c *
>
>: would. Is there a way to do this that is not too complicated, i.e. that can
>: be done in one command line?
>
>
>Peter
--
Herb Stein
The Herb Stein Group
www.herbstein.com
[EMAIL PROTECTED]
314 215-3584
------------------------------
From: [EMAIL PROTECTED] (Charlie Brown)
Crossposted-To:
alt.os.linux,comp.os.linux.hardware,comp.os.linux.setup,comp.sys.ibm.pc.hardware.storage
Subject: Re: HOT HD
Date: Sun, 23 Apr 2000 16:59:31 GMT
On Sun, 23 Apr 2000 14:51:32 GMT, [EMAIL PROTECTED] (Stewart
Honsberger) shocked the world by writing:
>On Sun, 23 Apr 2000 00:59:58 -0400, Edward M Grill wrote:
>> I have a Maxtor COMPUsa 10.2G 7200rpm ata/66 (only at 33) HD running
>>under Mandrake Linux 7.0 on a FIC VA 503+ motherboard and an AMD k6-2
>>450Mhz. I have successfully overclocked the CPU to 550 and a little more,
>
>Over-clocking is always dangerous, but unrelated to your situation.
>
>>currently only at 450 because I am afraid of a situation. even at 450mhz the
>>HD is EXTREMELY hot on its underside.
>
>Any HDD over 6GiB is reccomended to sit in a 5 1/4" bay with a HDD cooling
>unit. HDD coolers are relatively inexpensive, and could save you lots of
>trouble and expense somewhere down the road.
Who, exactly, is this recommended by?
Not the drive makers.
Possibly those who make the drive coolers?
Neither my 27 gig nor my 20 gig drives (each in a 3.5" bay) is
anywhere near overheating.
>
>--
>Stewart Honsberger (AKA Blackdeath) @ http://sprk.com/blackdeath/
>[EMAIL PROTECTED] (Remove 'thirteen' to reply privately)
>Humming along under SuSE 6.4, Linux 2.2.14
------------------------------
Reply-To: [EMAIL PROTECTED]
From: [EMAIL PROTECTED] (Rod Smith)
Subject: Re: Corel Linux
Date: Sun, 23 Apr 2000 17:07:04 GMT
[Posted and mailed]
In article <[EMAIL PROTECTED]>,
Koen Van Baelen <[EMAIL PROTECTED]> writes:
> Hi everybody,
>
> is Corel Linux any good?
That depends on your needs. For my thoughts on Corel Linux and several
other distributions, check my web site:
http://www.rodsbooks.com/distribs/
--
Rod Smith, [EMAIL PROTECTED]
http://www.rodsbooks.com
Author of books on Linux networking & multi-OS configuration
------------------------------
From: [EMAIL PROTECTED] (Donovan Rebbechi)
Crossposted-To: comp.os.linux.advocacy
Subject: templates and g++
Date: 23 Apr 2000 13:25:19 -0400
Does anyone here know how to get templates working with g++ ?
For example, I have a file graph.cpp and graph.h, and main.cpp.
It refuses to link unless I use #include "graph.cpp" in the file
main.cpp
excerpt from graph.h:
....
template<class DATA>
class Graph
{
public:
Graph();
......
}
excerpt from graph.cpp
template< class DATA >
Graph< DATA >::Graph ( )
{
...
}
from main.cpp:
#include "graph.h"
// If I uncomment this, I can't link because I get a complaint that
// Graph<int>::Graph(void) is "undefined"
// #include "graph.cpp"
//
int main()
{
Graph<int> G;
}
------------------------------
From: Leonard Evens <[EMAIL PROTECTED]>
Subject: Re: core dumps (or the lack of them)
Date: Sun, 23 Apr 2000 13:26:06 -0500
Stephen Charlton wrote:
>
> When I'm using most Linux distributions I'm given a core dump whenever
> there's a segmentation fault.
>
> However when using Slackware 7.0 every time I get a segmentation fault I
> don't get a core dump. Is there anything I need to change to produce
> core files? I need these core files for the debugging of my applications.
I don't know how Slackware is organized on this matter, but
generally a program will produce a core dump if the owner
of the corresponding process has permission to write in the
directory where the core dump would appear. Also, the shell
which starts the process may have limits set which affect
the maximum size of a core dump. If this is set at zero or
some very low number, no core dump will be produced. For
bash, the appropriate shell command is
ulimit -c N
where N is the maximum size allowed for a core dump or the term
unlimited.
--
Leonard Evens [EMAIL PROTECTED] 847-491-5537
Dept. of Mathematics, Northwestern Univ., Evanston, IL 60208
------------------------------
From: "Bigoffthewall" <[EMAIL PROTECTED]>
Subject: please help with network
Date: Sun, 23 Apr 2000 13:12:40 -0400
Ok i new to networking and dont really know where to begin but i have 2
puters running redhat 6.1. i can telnet to either puter but.. i wanna set it
up a router to serve my client its inet connection behind a firewall..I have
a firewall script written with ip chains and in /etc/rc.d/rc.init i have
these settings
echo 1 > /proc/sys/net/ipv4/ip-forward
echo 1 > /proc/sys/net/ipv4/ip_dynaddr
echo TRUE > /proc/sys/net/ipv4/ip_log_martians
echo TRUE > /proc/sys/net/ipv4/ip_addrmask_agent
echo FALSE > /proc/sys/net/ipv4/ip_source_route
echo FALSE > /proc/sys/net/ipv4/ip_accept_redirects
echo 1 > /proc/sys/net/ipv4/tcp_syncookies
i tried useing ppp0 to connect to the inet but it will not log on.now if use
kppp to dial out i have no problems loging on to my isp..When configuring
ppp under linuxconfig does it look to etc/resolve.conf for isp ip address or
do you configure it with it? My etc/reslove.conf looks like
search gloryroad.net
nameserver 216.228.97.50
nameserver 216.228.97.51
i have also tried
search trinity.net (my router)
nameserver 192.168.255.50
nameserver 216.228.97.50
nameserver 216.228.97.51
I have been told to use named or diald but i have no idea how to configure
either of them ..
Now for my client
im using a setting of
echo 1 > /proc/sys/net/ipv4/tcp_syncookies
i used linuxconf to set my default route to the ip i assighned my router
which is 192.168.255.50(should i be using a different address?)
since i dont have bind or diald i was told to keep my etc/resolve.conf to
search my isp so i kept it like
search gloryroad.net
nameserver 216.228.97.50
nameserver 216.228.97.51
i have also tried
search trinity.net (my router)
nameserver 192.168.255.50
but no luck.. at first i thought it may be my firewall but i took it out and
kept it from booting and still the same when i try to view web pages using
netscape it gives me the message cant find server..Any suggestions? am i
missing anything?
Any help would be greatly apprecitated..
Thanks in advanced
------------------------------
From: Tony Lawrence <[EMAIL PROTECTED]>
Subject: Freeedom to Innovate Network
Date: Sun, 23 Apr 2000 17:58:01 GMT
Please be sure to check out
http://www.microsoft.com/freedomtoinnovate/ which is (try
not to gag):
The FIN is a non-partisan, grassroots
network of citizens and
businesses who have a stake in the success
of Microsoft and the
high-tech industry.
Microsoft is, of course, well known for its rich history of
"innovation" ( I didn't look it up in the dictionary, but it
must now mean "buy, copy or steal somebody else's technology
and market the hell out of it").
This "grass roots network" (yeah, right) will help you write
your elected officials (Microsoft has already made them all
aware of how important this is by contributing to their
campaign funds) so that you can let them know how critical
it is that Microsoft be left alone to do whatever it damn
well pleases.
Remember that, according to Billy Boy, "Microsoft went the
extra mile to end this lawsuit through mediation" (funny, I
don't remember that). Bill is also apparently responsible
for the economy; he warns that we need to "protect the
innovation and freedom that has fueled the hight-tech
indistry, created millions of jobs and generated
unprecedented economic growth".
So yes, let's all join the "grass roots" Freedom To Innovate
Network.
--
Tony Lawrence ([EMAIL PROTECTED])
SCO/Linux articles, help, book reviews, tests,
job listings and more : http://www.pcunix.com
------------------------------
From: [EMAIL PROTECTED] (Dave Brown)
Subject: Re: linux programs, tools and functionality for windows?
Reply-To: [EMAIL PROTECTED]
Date: 23 Apr 2000 12:59:45 -0500
In article <mQDM4.114$[EMAIL PROTECTED]>, Peter Hutchison wrote:
>
>SAP User <[EMAIL PROTECTED]> wrote in message
>news:8dkhou$g35$[EMAIL PROTECTED]...
>> Hi there,
>>
>> due to a job change I have to work with windows instead of linux :-(((
>>
>> Are there any sites of information about programs for windows which
>> provide some of the functionality of a nice linux system?
>>
>> For example:
>> * bash
>
>AFAIK, there is only Doskey which allows for command history, editing and
>macros. Limted filename completion tho
>
>> * emacs
>I am sure there are some Win equivalent files around.,
>
>> * Hacks for the desktop to provide an extended functionality like in most
>...
>
>For more software files try:
>http://www.tucows.com
>http://www.shareware.com
>http://www.download.com
>http://www.simtel.com
>http://www.zdnet.com
There's a package around some of the archives called DosUnix (freeware),
which contains ports of a number of Unix utilities. There's version of
Vim for Win32. There used to (maybe still is) on Sunsite under "msdos"
a number of ports of Unix utilities. Especially useful is ZSH.EXE, which
is a port of the zsh, which is a fairly useful subset of bash/ksh, and
gives shell programming functions far beyond COMMAND.COM.
--
Dave Brown Austin, TX
------------------------------
From: Chris T <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.hardware,comp.sys.laptops,misc.forsale.computers.pc-specific.portables
Subject: FS: toshiba 430CDT and ZipDrive $480 and FDC settlement of $100!
Date: Sun, 23 Apr 2000 13:59:14 -0700
I have my Toshiba Satellite Pro 430CDT with an active matrix LCD for
sale. The CPU is a Pentium running at 120mhz, it has the CD-ROM
attachment and the Floppy Disk attachment. The condition is fair because
of a small crack near one of the hinges (I think this a common problem)
and there are a couple of stickers on the case (a CRYSTAL METHOD and a
SWITCH BLADE SYMPHONY). I still have the orignal manuals as I'm the
original owner. Also included is a PC-Card FaxModem rated at 56k Flex.
I will not erase the HD which has MS OFFICE 97 PRO, running MS Windows
98, MS MONEY 99 and more! You will be responsible to delete those Apps.
I will include a very nice laptop bag which the Satillite Pro fits very
well in and an external mouse (when you don't want to use the trackpoint)
NEW PRICE $480 (need the cash for taxes). Also as a recent addition I
will include my Zip Drive as well because I realized that I won't need it
after the sale. Also after going to Toshiba's web site to update the
drivers I notice a settlement from the FDC that entitles Toshiba laptops
owners towards $100 free on Toshiba products this is the web site for
more details:
http://www.csd.toshiba.com/cgi-bin/tais/support/notices/fdcsettlement.jsp
E-mail me for more info (one the laptop NOT the legal settlement I am not
a lawyer) or if you wish to buy the laptop at [EMAIL PROTECTED]
or I prefer [EMAIL PROTECTED]
------------------------------
From: "Justin R. Miller" <[EMAIL PROTECTED]>
Crossposted-To:
alt.os.linux,comp.os.linux.hardware,comp.os.linux.setup,comp.sys.ibm.pc.hardware.storage
Subject: Re: HOT HD
Date: Sun, 23 Apr 2000 18:15:33 GMT
Edward M Grill wrote:
> I have a Maxtor COMPUsa 10.2G 7200rpm ata/66 (only at 33) HD
Just FYI, at a web shop I once worked at we bought four of these same
drives (this was last summer). All ran extremely hot, even to the point
of one of them giving me a rather bad burn (of course this was partly my
fault :-) Anyway, that hot one died within a week and the other three
all died within a month, even though we tried to cool them moreso than
other drives (the case was even open). I would be careful with that
model, and perhaps all of Maxtor.
Just my $0.02,
Justin
--
_ ___ __ __
+-=====- _ | | _ \ \/ | -======- Justin R. Miller -==============-+
| | || | / |\/| | [EMAIL PROTECTED] |
+-=====- \__/|_|_\_| |_| -======- http://openup.com/justin/ -=====-+
------------------------------
Date: Sun, 23 Apr 2000 18:26:46 +0000
From: mh <[EMAIL PROTECTED]>
Subject: EMERGENCY!!!
I'm facing the possible loss of nearly 10 years worth of data and
assorted applications. Here's my situation:
I use a secondary (SCSI) hard drive (sdb) to make a copy of all data and
application files stored on the primary hard drive (sda). I have no tape
backup, nor any other form of backup. Both drives are IBM 9.1 GB.
I decided I needed to re-partition my primary drive (sda), and wanted to
upgrade the kernel as well. I umounted the secondary drive (sdb) and
proceeded to perform a clean install of RH 6.0 on sda. I repartitioned
sda using disk druid, and was careful to deselect sdb under the valid
drives option as I created each partition.
After completing the install, I attempted to mount sdb to copy the data
and files back to sda. I got an error message saying there was a
problem with the filesystem, superblock, or there were too many
filesystems mounted.
I ran "cfdisk /dev/sdb" and cfdisk reported the entire drive (single
partition) as "free space"!! I ran "e2fsck /dev/sdb" and it reports
problems with several inodes and prompts me to "relocate". I am afraid
to do this, since I fear losing ALL data and have no other backup.
Unless disk druid did something to sdb, the drive should be intact. I'm
hoping the errors have something to do with the way the extended
filesystem/super block was written on sda rather than a problem with
sdb.
Suggestions/recommendations will be much appreciated.
------------------------------
From: Se�n � Donnchadha <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.ms-windows.nt.advocacy,comp.os.ms-windows.advocacy,comp.lang.java.advocacy
Subject: Re: How Microsoft inhibits competition & innovation
Date: Sun, 23 Apr 2000 14:27:07 -0400
[EMAIL PROTECTED] (Roger Blake) wrote:
>>
>>Like I said, those OSs don't have nearly as many hardware
>>manufacturers writing drivers for them. And those that do can afford
>
>You're reaching here, trying to blame Windows' notorious problems
>on drivers from hardware vendors.
>
Hey, it's just my experience, that's all. Microsoft bashers talk on
and on about Windows' "notorious problems", but when asked to drill
down on the vast majority of them, they come up with squat. A great
example of this is DLL Hell. In discussions like this one, I've asked
dozens of people moaning about DLL Hell to list just a few problems
they've actually traced to DLL conflicts, and apart from half a
handful of well-known instances where the vendor (usually Microsoft)
screwed up a specific DLL upgrade, their responses always add up to a
big fat zero. The same goes for registry corruption, OS-level memory
leaks, etc. People making these claims usually hate Windows to begin
with and don't bother researching problems when they occur. To their
pet OSs, of course, they give much more slack. Another point worth
mentioning is that when talking about these issues, advocates
invariably leave out the fact that non-Microsoft OSs are just as
vulnerable as Windows to most of them. Do you really think Unix/Linux
is immune to shared library conflicts or misbehaving drivers?
>>
>>What design issues?
>
>There are so many deficiencies with Windows design that it's hard
>to find a place to start! On a high level, one good example is the
>braindead design of the registry which in its Win95 incarnation would
>grow boundlessly until it wouldn't permit the OS to boot up!
>
The registry does seem to never shrink on Win9x, but I've *NEVER* seen
it prevent the OS from booting, and I've used Win9x since the Chicago
beta days. And after 10 years of Unix development, I welcome the
concept of the registry with open arms.
>
>It's still
>so damned fragile and nearly impossible to debug that now Windows 98
>keeps multiple copies around and tries to heal itself if corruption
>is detected.
>
No, it just keeps several backups around for disaster recovery (it
doesn't try to "heal itself"). What the heck's wrong with that?
>
>Not to mention that so many registry entries use the
>short version of a file's name that, given the way long filenames are
>implemented, you can't even make a reliable backup of the OS if your
>backup software works through the filesystem!
>
Since the registry doesn't know a filename from any other kind of text
string, that would be an application problem.
>
>Then how about a network subsystem whose design dictates that the
>response to network errors is that THE ENTIRE SYSTEM HANGS INTERMINABLY.
>
How about that's a pile of crap? I've never experienced that, nor even
heard of it. Now that you mention network errors though, I do remember
working for a couple of Unix ISVs that provided diskless Sun
workstations to its employees, hooked up to big Sun servers. And I do
remember staring for hours at repeating "NFS server <name> not
responding" messages.
>
>Another issue is the endless problems with shared DLLs. EVEN DLLs
>PRODUCED BY MICROSOFT ARE FREQUENTLY INCOMPATIBLE ACROSS VERSIONS AND
>CAN CAUSE SEVERE SYSTEM PROBLEMS!
>
Yes, Microsoft has screwed up DLL upgrades several times. So what?
They're just shared library conflicts. Unix is just as susceptible to
them. If you really think that Unix's convention of using symlinks and
embedding versions into shared library filenames makes it immune, then
you don't really understand the problem.
>
>For that matter, what other OS has a design that practically requires
>non-trivial applications to modify the operating system in order to run?
>
You've got it backwards. Windows does *NOT* have a design that
requires nontrivial apps to modify the OS. However, many apps *DO*
modify the OS anyway. Why? Because of the intensity of commercial
competition. Each Windows ISV is constantly trying to figure out ways
to make their app do something the competition can't do - something
"magic" - and the only way to do that is (by definition) to modify the
OS. Examples of this are things like ZIP utilities that weld
themselves into the file system and desktop tools that take over the
OS's entire GUI. Then there are the ISVs that simply don't follow the
rules - not Microsoft's rules, but common sense rules. For example, I
still can't figure out why Intuit apps put stuff into the Windows
directory. Nevertheless, my point is supported by the fact in those
software areas where there's less competition (or because of the
nature of the software), there's always less hanky-panky. For example,
Windows games usually don't touch the OS, nor does a huge amount of
nontrivial freeware and shareware.
>
>In Linux, to install an app, I copy the files to a directory and run
>the executable. If anything, the only other thing that may be needed
>is to install a support library. That's it. To uninstall, just
>delete the files! Amazing!
>
Oh, you're leaving out a *BUNCH* of stuff here. In fact, Unix is way
*WORSE* in this regard than Windows. That's because nontrivial Unix
apps have to sprinkle bits and pieces of themselves all over the file
system. Let's see now. The executable or launcher goes into one of the
"bin" directories, the data goes into one of the "lib" directories,
global settings go God knows where, per-user settings go into home
directory dotfiles, log files go into one of the "etc" directories or
some place under "/var", and heaven help you if the app installer
wants to give you some way to access the app from the GUI. And that's
just for relatively simple apps. Just wait until commercial ISVs start
giving your stable little OS their "magic" treatment.
>
>One of the best-known
>of these was moving display functions into the unprotected "Ring 0"
>of the OS -- now a buggy display driver can crash your enterprise
>server.
>
Now this is one anti-Microsoft argument that never fails to amaze me.
A device driver can crash the OS?! Yeah, like Unix is safe from bad
device drivers. And as for display functions being in "Ring 0", let me
remind you that one of the most reliable microcomputer Unixes in
history was SunOS 4.x, with the SunTools GUI in the kernel. It took
many years for Solaris to become that stable (and many would argue
that it still isn't) after they made the switch to SVR4.
>>
>>For God's sake, why don't you read what was said before replying to
>>it? It's not Microsoft that creates the device drivers; it's the
>
>I would submit that Windows still has a propensity to crash due
>to shared DLL problems, registry problems, and a host of internal
>design problems (including 16-bit code in Win9x that can easily
>hang the entire system if a user application crashes). This would
>still occur even if only generic, MS-supplied drivers were used.
>
I strongly disagree. Most of the device drivers MS supplies weren't
developed by MS. To really test the quality of the OS's core, you'd
need to run it in a very unusual configuration - plain VGA, plain IDE,
no network, etc. I doubt very much that you've spent enough time with
Windows in such a configuration to make the above claim.
>>
>>Come on. Mac reliability is a joke, and the reason it may not be as
>
>I have not found this to be the case. We use PowerMac systems to do
>digital video and they have been quite reliable in running the demanding
>applications involved. I couldn't even imagine trying to do anything
>similar on Windows systems.
>
It doesn't matter if you can't imagine it. The fact is that it's
happening everyday. Millions of people use Windows to run demanding
applications. And let's not forget that Mac reliability is almost by
definition a joke compared to that of WinNT/Win2K.
>
>The general public (the largest group of MS Windows users) clearly
>believes that Microsoft is responsible for most (if not all) advances
>in computing. Get out from behind your keyboard once in a while and talk
>to users, you'll see this is the case.
>
I have talked to users, and I don't see that this is the case at all.
Besides, even if it were the case, it isn't exactly Microsoft's
responsibility to educate consumers in computer science, nor is it
their job to inform consumers of the choices they have.
>
>[standard anti-Microsoft babbling deleted]
>
What can I say? I totally disagree. In my experience, Microsoft's
stuff is on average of *BETTER* quality than the majority of the
others. Some of their design decisions were clearly motivated by
marketing considerations rather than the desire to create "beautiful
technology", but I don't think it's fair to fault them for that. They
are, after all, not a nonprofit research organization.
------------------------------
** 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.misc) 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-Misc Digest
******************************
