Linux-Misc Digest #603, Volume #25 Mon, 28 Aug 00 23:13:02 EDT
Contents:
Re: kernel compiles root device as (3,65); boots for (3,41) (Dirk Foersterling)
Difference between 'Mail' and 'mail'? (softrat`)
Re: Linux spontaneously changes time (Hal Burgiss)
Re: X-Window must die! What's alternative? ("Andrew N. McGuire ")
lp failure (Russ Button)
Re: Pro*C (Jean-David Beyer-valinux)
Re: dump win98 partition (Robert Jones)
Re: print control (Garry Knight)
Re: Amateur Hacker Backdoors Thwarted By Upgrade? (Jean-David Beyer-valinux)
Re: Graphs program ! (Robert Love)
Re: Linux, XML, and assalting Windows (Bob Hauck)
Re: Linux, XML, and assalting Windows (Bob Hauck)
How to make NumLock behave on my ThinkPad in KDE/Gnome? (Jerome Mrozak)
Re: Linux spontaneously changes time (Jean-David Beyer-valinux)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Dirk Foersterling)
Crossposted-To: comp.os.linux.development.system
Subject: Re: kernel compiles root device as (3,65); boots for (3,41)
Date: 28 Aug 2000 04:40:40 GMT
On Mon, 28 Aug 2000 01:29:24 GMT, xavian anderson macpherson wrote:
>
>M. Buchenrieder wrote:
>>
>> man rdev
>> man lilo.conf
>>
>> Michael
>> --
>> Michael Buchenrieder * [EMAIL PROTECTED] *
>http://www.muc.de/~mibu
>> Lumber Cartel Unit #456 (TINLC) & Official Netscum
>> Note: If you want me to send you email, don't munge your address.
>--------------------------------------------------------------------------
>I HAD ALREADY READ THE MAN RDEV. IT DID NOT TELL ME HOW TO SET THE DEVICE
>FROM 3,41 TO 3,65 OR REVERSE. if you know the exact code to fix this, give
First, it does. Here's the passage:
When using the rdev, or swapdev commands, the root_device
or swap_device parameter are as follows:
/dev/hda[1-8]
/dev/hdb[1-8]
/dev/sda[1-8]
/dev/sdb[1-8]
If you look at /dev/hdb1 using ls -l, you will see something like that:
brw------- 1 root root 3, 65 Nov 15 1993 /dev/hdb1
This is Device 3,65. The other number you mentioned is 3,41. This would
be /dev/hda41 which I don't think is correct. Are you sure that you
didn't get confused by hex numbers? (Hex 41 = Dec 65).
-dirk
--
D i r k F "o r s t e r l i n g
[EMAIL PROTECTED] ******** http://www.DeathsDoor.com/milliByte/
-------------
"If p gleich tail then wird hilf auch auf tail gesetzt." - R.K.
------------------------------
From: softrat` <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux
Subject: Difference between 'Mail' and 'mail'?
Date: Mon, 28 Aug 2000 19:11:23 -0700
What is the difference between the user directories 'Mail' and 'mail'?
What generates them? May they be combined? How? Thank you for your help.
the softrat
[EMAIL PROTECTED]
--
"I've got some amyls. We could either party later or, like, start his
heart."
-- "Cheech and Chong's Next Movie"
------------------------------
From: [EMAIL PROTECTED] (Hal Burgiss)
Crossposted-To: comp.os.linux.questions
Subject: Re: Linux spontaneously changes time
Reply-To: Hal Burgiss <[EMAIL PROTECTED]>
Date: Tue, 29 Aug 2000 02:15:34 GMT
On 28 Aug 2000 23:48:17 GMT, David Rysdam <[EMAIL PROTECTED]> wrote:
>Mine is doing almost exactly the same thing. Here is more data:
>
>-No Java
>-Connected to LAN (I reset the time using rdate)
>-Happens at night only (work machine)
>-Intermittent problem (happens for a coupleof weeks, then stops
>happening for a few months)
>-I haven't noticed if it is always X hours
>
>>Would anyone care to guess what on earth is resetting my system clock,
>>or even where I would begin to look? Thanks in advance,
Blast from the past, and my mind is blurry. But I vaguely remember
something about changes in where the timezone info was stored. Some
libc5 apps would look in the wrong place, and then screw the clock just
as you describe. asclock was one that did this IIRC. Sorry I don't
remember details. Maybe search deja, google, or redhat.com. This came up
several times on redhat mailing list about the time of 6.1 (or was it
6.0?).
--
Hal B
[EMAIL PROTECTED]
[EMAIL PROTECTED]
--
------------------------------
From: "Andrew N. McGuire " <[EMAIL PROTECTED]>
Subject: Re: X-Window must die! What's alternative?
Date: Mon, 28 Aug 2000 21:13:50 -0500
On Tue, 29 Aug 2000, Bob Fahey quoth:
~~ Date: Tue, 29 Aug 2000 01:04:34 GMT
~~ From: Bob Fahey <[EMAIL PROTECTED]>
~~ Newsgroups: comp.os.linux.misc
~~ Subject: Re: X-Window must die! What's alternative?
~~
~~
~~ "Hans" <[EMAIL PROTECTED]> wrote in message
~~ news:8lmqqa$isv$[EMAIL PROTECTED]...
~~ >
~~ >
~~ > Hello there,
~~ >
~~ > I read an article "X must die." in www.linux.com.
~~ > Is there X-Window alternative?
~~ >
~~ > Cause KDE or GNOME environment is not better than compared with M$
~~ > Windows 9x. They are the clones of M$ Windows 9x. At least Pentium II
~~ > with 128M ram manages KDE application smoothly in my experience. I feel
~~ > Linux in GUI environment seems to go backwards. For example, when I do
~~ > 'startx', run Netscape then I suddenly return 1995 before using Windows
~~ > 95.
~~ >
~~ > Using Linux in text mode is very nice. I like it. Its Kernel is compact,
~~ > fast, provides manys ways of doing a job (Network, Local, Cluster.)
~~ > under low cost. :)
~~ >
~~ > Is there X-Window alternative?
~~
~~ It's called, "CONSOLE" you bag o' rocks.
~~ I sometimes wonder how this world even turns on its axis carrying around
~~ dead weight like this.
[ snip sig ]
~~ Have fun with the most flexible OS in the world. Linux. It even serves up
~~ spoiled rotten users like you!
~~ Bob sends
You don't think that was a bit uncalled for?
Granted his post was inaccurate, but there is not
need to resort to name calling!
anm
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~ Andrew N. McGuire ~
~ [EMAIL PROTECTED] ~
~ "Plan to throw one away; you will, anyhow." - Frederick P. Brooks, Jr. ~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
------------------------------
From: Russ Button <[EMAIL PROTECTED]>
Subject: lp failure
Date: 29 Aug 2000 02:14:11 GMT
My host is a PC running Mandrake 2.2.13-7. But the problem is
really a lpd issue I think. As you might have guessed, I'm
unable to print. The printer is an HP network printer.
An HP Laserjet 4050N. Here's my printcap entry for it.
lp:\
:sd=/var/spool/lpd/lp:\
:mx#0:\
:sh:\
:rm=172.19.0.5:\
:rp=lp:\
:if=/var/spool/lpd/lp/filter:
My machine is getting IP address, etc, from a dhcp server.
The hostname for the machine is "bags", which is configured
through linuxconf. When I attempt to lpr a simple text
file the cursor just returns and then... nothing. The
only interesting clue is when I run lpq I get:
bags>lpq
Warning: unable to get official name for local machine bags
no entries
My hosts file has no entry for bags of course as the machine
gets its IP addr from dhcp.
Any suggestions as to what might be wrong?
Much thanks!
Russ
A Solaris guy trying to adjust to a brave new world.
------------------------------
From: Jean-David Beyer-valinux <[EMAIL PROTECTED]>
Subject: Re: Pro*C
Date: Mon, 28 Aug 2000 22:19:51 -0400
mst wrote:
> [EMAIL PROTECTED] wrote:
> >
> > cpp is the C++ compiler, Pro*C is a precompiler that does embedded sql
> > code for c and cpp
> >
>
> Maybe in windoze it is, but in Linux, cpp is the "c pre-processor". The
> C++ compiler is g++. Check the man page for cpp. I don't know if it will
> work with the sql code.
>
> MST
cpp and g++ do not handle SQL directly. What I do is have makefiles that
look like this (in part):
.cc.o:
${CXX} -c ${CXX_FLAGS} ${CXX_INCLUDES} $<
.C.o:
${CXX} -c ${CXX_FLAGS} ${CXX_INCLUDES} -I${IBM_DB2_HEADERS} $<
.sqC.C:
${IBM_DB2_DB2} CONNECT TO stock
${IBM_DB2_DB2} PREP $< DEGREE 1 ISOLATION RS PREPROCESSOR \"g++ -x c++ -o
$*.i -E -P ${CXX_INCLUDES} -I${IBM_DB2_HEADERS}\" bindfile
${IBM_DB2_DB2} BIND $*.bnd
${IBM_DB2_DB2} DISCONNECT stock
${IBM_DB2_DB2} TERMINATE
This is for the IBM db2 dbms that has embedded SQL in the C++ programs.
Those programs have suffixes .sqC. The third bunch of stuff talks to db2
and executes commands in that. It runs the C++ preprocessor and one of its
own (I forget in which order), and makes a file that the C++ can read with
suffix .C. The second bunch of stuff can convert .C files into .o files.
The first bunch is for my normal C++ files that have suffix .cc. I am sure
IBM would not have put this capability into their db2 programif the C++
compiler could do it all by itself. But the C++ standard does not have
anything about SQL in it, and it never sees the SQL statements.
It turns a C++ program with embedded SQL statements that looks something
like this:
EXEC SQL
DECLARE ugCursor CURSOR FOR
SELECT I.ibd_company_id, T.tick_symbol,
I.ibd_eps, I.ibd_rs, I.ibd_acc_dst
FROM vl_ranks V, ibd I, tick T
WHERE V.vl_timeliness = 1
AND V.vl_date = :ad_f_date
AND :ad_f_date <= I.ibd_date
AND I.ibd_date < :ad_l_date
AND I.ibd_company_id = V.vl_company_id
AND I.ibd_company_id = T.tick_company_id
AND T.tick_f_date <= :ad_f_date
AND :ad_f_date <= T.tick_l_date
ORDER BY I.ibd_eps DESC, I.ibd_rs DESC, I.ibd_acc_dst
FOR READ ONLY
OPTIMIZE FOR 5 ROWS;
into something like this:
{
#line 10926 "ugDerived.i"
sqlastrt(sqla_program_id, &sqla_rtinfo, &sqlca);
#line 10926 "ugDerived.i"
sqlaaloc(2,5,2,0L);
{
struct sqla_setd_list sql_setdlist[5];
#line 10926 "ugDerived.i"
sql_setdlist[0].sqltype = 460; sql_setdlist[0].sqllen = 11;
#line 10926 "ugDerived.i"
sql_setdlist[0].sqldata = (void*)ad_f_date;
#line 10926 "ugDerived.i"
sql_setdlist[0].sqlind = 0L;
#line 10926 "ugDerived.i"
sql_setdlist[1].sqltype = 460; sql_setdlist[1].sqllen = 11;
#line 10926 "ugDerived.i"
sql_setdlist[1].sqldata = (void*)ad_f_date;
#line 10926 "ugDerived.i"
sql_setdlist[1].sqlind = 0L;
#line 10926 "ugDerived.i"
sql_setdlist[2].sqltype = 460; sql_setdlist[2].sqllen = 11;
#line 10926 "ugDerived.i"
sql_setdlist[2].sqldata = (void*)ad_l_date;
#line 10926 "ugDerived.i"
sql_setdlist[2].sqlind = 0L;
#line 10926 "ugDerived.i"
sql_setdlist[3].sqltype = 460; sql_setdlist[3].sqllen = 11;
#line 10926 "ugDerived.i"
sql_setdlist[3].sqldata = (void*)ad_f_date;
#line 10926 "ugDerived.i"
sql_setdlist[3].sqlind = 0L;
#line 10926 "ugDerived.i"
sql_setdlist[4].sqltype = 460; sql_setdlist[4].sqllen = 11;
#line 10926 "ugDerived.i"
sql_setdlist[4].sqldata = (void*)ad_f_date;
#line 10926 "ugDerived.i"
sql_setdlist[4].sqlind = 0L;
#line 10926 "ugDerived.i"
sqlasetd(2,0,5,sql_setdlist,0L);
}
#line 10926 "ugDerived.i"
sqlacall((unsigned short)26,1,2,0,0L);
#line 10926 "ugDerived.i"
sqlastop(0L);
}
which, though certainly obscure, is valid C++ code that gets the job done.
--
.~. Jean-David Beyer Registered Linux User 85642.
/V\ Registered Machine 73926.
/( )\ Shrewsbury, New Jersey
^^-^^ 9:05pm up 20 days, 4:34, 1 user, load average: 0.99, 0.97, 0.91
------------------------------
From: Robert Jones <[EMAIL PROTECTED]>
Subject: Re: dump win98 partition
Date: Mon, 28 Aug 2000 21:26:52 -0500
Dances With Crows wrote:
> On Mon, 28 Aug 2000 20:21:52 -0500, Robert Jones wrote:
> >Please let me pose another semi-hypothetical case that I haven't tried yet
> >simply because I don't have a spare hard drive and I haven't destroyed my
> >running Linux drive. (Yet).
> >I'm running RH6.0 in a dual boot configuration. Win95 is on hda; Linux on hdc,
> >partitioned as follows
> >hdc1 boot
> >hdc5 swap
> >hdc6 root
> >I make full backups regularly using a script that includes:
> >tar -cvvf /dev/nst0 -M / -X /home/rj/nobackup -V "Full Backup $ratnow"
> >($ratnow contains the date/time string; nobackup excludes /dos, /proc and
> >~/.netscape/cache)
> >O.K., so I have a 2-tape set of everything I need to completely
> >overwrite a running system. *IF* I have a running system. Suppose the
> >worst happens: My hdc drive turns into a pile of scrap metal, I replace
> >it, boot the rescue system from floppy and use fdisk, mke2fs and mkswap
> >to make my new drive resemble the old one in better times.
> >
> >After I use mknod to get the rescue system acquainted with /dev/st0 and
> >/dev/nst0, can I proceed to:
> >$ mkdir /newmnt # (On the ramdisk)
> >$ mount -t ext2 /dev/hdc6 /newmnt # mount the root drive
> >$ mkdir /newmnt/boot # create a mount point for the boot partition
> >$ mount -t ext2 /dev/hdc1 /newmnt/boot
> >$ cd /newmnt
> >$ tar -xvvf /dev/nst0 -M
> >...and restore everything in one fell swoop, or am I asking for too much?
> >If that will work, my friends who run Win9x will surely turn green!
>
> You'll have to run LILO again, because tar doesn't restore files to the
> exact places on disk that they were at before, and LILO needs to know
> the exact disk blocks that the kernel/loading map occupy. This can be
> done rather easily, right after that last tar command:
>
> # chroot /newmnt /sbin/lilo
>
> (This is roughly similar to doing "SYS C:" from DOS/9x, except much
> better, of course.)
Yessir, and the whole thing is a dang sight mo' betta that having to reinstall
Win95 in order to access the nice GUI that was used to backup Win95 so that Win95
can be restored from tape. [eyes roll uncontrollably]
Thanks for the super quick reply!
> Reboot, et voil�!
Regards,
Robert
------------------------------
From: Garry Knight <[EMAIL PROTECTED]>
Subject: Re: print control
Date: Tue, 29 Aug 2000 03:18:47 +0100
[EMAIL PROTECTED] wrote:
>Hi
>
>Suppose I am printing a 10-page text file. After printing is finished, I
>realize there is a typo, on page 4. Now, rather than printing the whole
>document again, is there any way to print only page 4?
Using which particular software?
--
Garry Knight
[EMAIL PROTECTED]
------------------------------
From: Jean-David Beyer-valinux <[EMAIL PROTECTED]>
Subject: Re: Amateur Hacker Backdoors Thwarted By Upgrade?
Date: Mon, 28 Aug 2000 22:32:29 -0400
I R A Darth Aggie wrote:
> On Mon, 28 Aug 2000 12:52:19 -0400,
> Wretch <[EMAIL PROTECTED]>, in
> <[EMAIL PROTECTED]> wrote:
>
> + However, in my post I was sort of groping, in an offhanded way,
> + for some insight as to exactly how one installs a "backdoor"
> + once they get into the system. Can this be done in an effectively
> + infinite number of ways, or are there typical tricks that I might
> + be able to look for?
>
> Yes and yes. The former makes the later unreliable. Thus wipe-n-load is
> the way to go. That and:
>
> # cat /etc/hosts.allow
> ALL: ALL
>
> tcpwrappers are your friends.
They are, but does the above help anyone but a hacker (in the bad sense)?
I assume you meant to suggest that /etc/hosts.deny say:
ALL: ALL
and that /etc/hosts.allow say relatively little. For a long time, I got by
with:
ALL : LOCAL
but since networking my two machines together, I have added:
# touchl is 192.168.201.1
portmap : 192.168.201.1
time : touchl
where touchl is the name of my other machine, and 192.168.201.0 is my LAN.
I infer that ssh, nfs, and samba do not use tcpwrappers, at least with
inetd.
--
.~. Jean-David Beyer Registered Linux User 85642.
/V\ Registered Machine 73926.
/( )\ Shrewsbury, New Jersey
^^-^^ 9:05pm up 20 days, 4:34, 1 user, load average: 0.99, 0.97, 0.91
------------------------------
From: Robert Love <[EMAIL PROTECTED]>
Subject: Re: Graphs program !
Date: 28 Aug 2000 21:36:51 -0500
>>>>> "Henning" == Henning <[EMAIL PROTECTED]> writes:
Henning> Hi all, I'm looking for a good graphs/math program for
Henning> linux. I have tried Gnuplot but I find it to be less
Henning> than desirable. I would like it to have a GUI. And as
Henning> the output is to be used in latex docs, I would like it
Henning> to be able to export to a format readable be latex.
Do you require free programs? If you're willing to pay for
them try either Mathematica or Macsyma. Both are extremely
capable math and graphing programs and both are available for
Linux although the cost can be high if you're not a student.
For simple graphs with high quality output I would stick to
Gnuplot and try one of the front ends to it. I've used Gnuplot
in my TeX documents with no problem.
--
=============================================================
| Support Signature Minimalism |
=============================================================
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: alt.os.linux,comp.text.xml,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Linux, XML, and assalting Windows
Reply-To: bobh{at}haucks{dot}org
Date: Tue, 29 Aug 2000 02:43:48 GMT
On Mon, 28 Aug 2000 02:28:10 GMT, paul snow <[EMAIL PROTECTED]> wrote:
>Bob Hauck <[EMAIL PROTECTED]> wrote in message
>news:[EMAIL PROTECTED]...
>> You've got it exactly backwards. Raw storage is just numbered blocks
>> on the disk. Filesystems are an abstraction created by the OS.
>No, you have it backwards. Where is the OS when your computer is off?
In a pile of bits on the hard disk.
>You turn your computer on. Where does the OS come from?
The boot loader reads it from the disk. The boot loader does not
necessarily know anything about the filesystem on the disk, although
some do. It just needs to know what block to start reading at, how
many blocks to read, and what address to jump to after the OS kernel is
loaded.
>In fact, given the same file system (no matter how it was
>constructed), you get the same behavior once the computer is turned on.
Which may or may not be what you want, since the identical disk could
be installed in a completely different computer. So now your tool has
to probe for hardware and the like too, unless the human tells it that
information.
Why are we building this again, instead of letting the OS do the work
for us?
>That is because your file system is nothing more than a persistent data
>structure. Nothing more. No magic.
It is nothing but a pile of bits until some software interprets it.
The persistent data structures are just patterns of bits that only have
meaning in the context of some software interpreting them. The boot
process is called "bootstrapping" for a reason.
On reset, the CPU hardware jumps to a reset vector in ROM and starts
executing what it finds as instructions. The ROM code causes the
hardware to read a boot block from disk. The boot block loads a kernel
from a known location (block) of the disk. The kernel loads drivers
and so forth and the system is running. I've simplfied and left many
details out out, but the point is that the software leverages it's way
step-by-step from a very simple understanding of what the bits on the
disk are (numbered blocks or similar that contain code to run) to a
state where they can be interpreted as files and directories. There is
no "chicken and egg" problem, just one piece of code loading another in
a carefully-orchestrated sequence. And the supposed data structures on
the disk do not drive the process until it is well along.
>But your file system is also the data structure that defines your OS
>and its applications. All the abstractions come into existence only
>after your software is loaded into memory from these data structures,
>and your software begins to run.
There *are no* data structures on a disk. There are blocks of bits.
Data structures are an abstraction created by software. Filesystems
are data structures. Different software (BIOS, boot loader, OS) can
and do treat the same data in different ways.
Filesystems as they live on disk are not self-describing. Sure, tools
besides operating system utilities can create filesystems and even read
and write them, but there is nothing there but patterns of bits until
some software somewhere interprets the bits.
This is why tools like Ghost can typically only be used to create exact
copies of a system. The tools don't know anything about the logical
structure of the data they manipulate. Don't get me wrong here, they
*could* know, and be able to manipulate the files "behind the back" of
the OS. That's been done. For example, you can take a DOS disk image,
mount it under Linux using the loopback device, edit it, and then write
it to a real disk. I can see that as something useful if you do mass
rollouts of identical systems, which is why people buy tools like
Ghost, but it seems cumbersome as a standard way to install software.
Your idea as I understand it is that one could have "generic" code to
do this manipulation of filesystems. You would describe the various
filesystem structures and things like the registry with XML-based
databases. You would have other databases that describe installation
requirements of your software, and a generic installer that does the
right thing based on these databases. All that is surely possible,
with a lot of effort, but I fail to see how the problem is simplfied by
doing that. You are essentially creating a specialized programming
language, but you seem to want to get away from programming.
On top of that, from a practical perspective, managing software from
outside the OS would seem to require rebooting the computer every time
you wanted to make a change. That is a step backwards for nearly all
server operating systems.
--
-| Bob Hauck
-| To Whom You Are Speaking
-| http://www.haucks.org/
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: alt.os.linux,comp.text.xml,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Linux, XML, and assalting Windows
Reply-To: bobh{at}haucks{dot}org
Date: Tue, 29 Aug 2000 02:43:51 GMT
On Mon, 28 Aug 2000 04:06:33 GMT, paul snow <[EMAIL PROTECTED]> wrote:
>> My response is in the other part of the thread. I'm not going to
>> repeat it here. Basically, it sounds like you want some sort of
>> meta-language to define installation procedures. This meta-language
>> would generate install programs for each supported platform, or perhaps
>> a database of some kind that could be used by the universal installer
>> to actually do the install.
>>
>> You are aware that this is sort of how Installshield and RPM work,
>> right? The developer creates scripts that describe his installation
>> and the tool makes some assumptions, and everything usually works. You
>> just want to make these scripts more abstract so that they'll work on
>> different platforms, and make them editable so you can pre-configure
>> your local setup.
>>
>> Is that about right?
>
>Yes, I am quite aware that this is how Installshield and RPM work. That is
>why I claim that all the information needed to re-factor the problem is
>right there in our hands already.
>
>Just stop for a moment and focus on the problem. Forget the execution
>environment.
But 'correctness' of a software installation is in reference to a
particular execution environment.
>From a data structure point of view, who really cares if one file is a
>font file, one is executable, another is an initial database, a
>library,
Existing installation tools don't care. I agree that it doesn't matter
from the point of view of the installation. It might be helpful when
you are trying to troubleshoot to know what files are for though.
>So of course I don't want to generate install programs. Why would I?
So you can write one generic one and apply it to many different
platforms. I thought that's what you were after.
>Take the XML descriptions of what structures are required for each
>software component. Take any set of options that are givens for this
>configuration. Evaluate what should then be done to construct each
>software component into this data structure that happens to be a file
>system. Then just render the proper structures into storage, and keep
>track of what structures are being modified, and why.
Ok, you're glossing over a *lot* here, but let's go with that. You're
talking about modifying the existing data structures in order to
"render" your application into the proper places. Which means that
your tool has to know what the structures are and how to manipulate
them in ways that the OS will treat as valid (your program does, after
all, have to run). Which means you need some language for describing
the structures, which you will presumably build using XML. This sounds
a lot like writing install programs to me. You're just moving the
complexity from code to data.
>If this is being done outside the OS (logically or even literally)
>then there is really no reason the same facility can't manage
>different platforms (like Solaris, Windows, Palm pilots, Linux, etc.).
Well, on NT, Solaris and Linux it would _have_ to be done literally
outside the OS, as those don't take too kindly to changing the
structures out from underneath the OS while it is running. Bad things
happen when you do that.
>Do I get some advantages? You bet. I can manage an OS even if the
>OS isn't functional.
Yes, but on the flip side it _can't_ be functional while you are
managing it. That's a step backwards.
>I can detect if structures that should be locked down have been
>changed.
You mean, your existing tools can't tell you this? Mine can.
>If I involve the various Operating Systems in this process, and all
>the abstractions they define, then I complicate the process.
Only because you have glossed over a huge number of implementation
details by simply assuming that "data structures" are simpler than
"code". You can move the complexity back and forth between the two,
but that does not make it go away.
--
-| Bob Hauck
-| To Whom You Are Speaking
-| http://www.haucks.org/
------------------------------
From: Jerome Mrozak <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.x,comp.windows.x.kde,comp.os.linux.setup,comp.sys.laptops
Subject: How to make NumLock behave on my ThinkPad in KDE/Gnome?
Date: Mon, 28 Aug 2000 21:55:19 -0500
I have an IBM ThinkPad 1412 with Mandrake 7.1 installed. My problem
happens under Gnome and KDE, but I'll concentrate on KDE.
My NumLock key is accessed through Shift+ScrollLock, but it isn't being
done right under the X sessions. This is a problem for me because
various letter keys share the numlock functions ("u" or "4", etc.) and
my "ls -A" becomes "3s -A".
I have a graphical login, and when there the NumLock light stays OFF.
This is GOOD. During the login process the NumLock light turns ON.
This is BAD.
I can press Shift+ScrollLock and I hear a beep from the unit. The
displayed keys then change to (or from) NumLock mode. But that NumLock
light stays lit.
I've examined the various ~/.* files (.Xinits) for the user's home
directory. I've monkeyed with /etc/X11/.Xmodmap and XF86Config, but
things don't really get better. In fact, things get pretty funky, with
the keyboard shifting between cursor modes and number modes, but not
back to alpha modes.
So...can I get a short lesson on which X or KDE files I should use to
affect numlock? I suspect them because NumLock works OK in text mode.
Thanks,
Jerome.
--
Jerome Mrozak "Never buy a dog and bark for yourself"
[EMAIL PROTECTED] --"Slippery" Jim DiGriz
(the Stainless Steel Rat)
------------------------------
From: Jean-David Beyer-valinux <[EMAIL PROTECTED]>
Subject: Re: Linux spontaneously changes time
Date: Mon, 28 Aug 2000 22:57:55 -0400
Christopher Wong wrote:
> One of my Red Hat 6.2 systems is spontaneously changing its system
> clock. I'm quite puzzled at this, since it is hardly running anything,
> not even crond. The only major app running is Sun's Java runtime
> (1.2.2). Here are some clues:
>
> 1. Always changes by -4hrs. My local time zone is EST5EDT. So it looks
> timezone-related. It used to change by -5hrs when my hardware clock
> was off by an hour.
>
> 2. Hardware real-time clock is unchanged (set to local time).
>
> 3. My Java app calls Runtime.currentTimeMillis(), for anyone who is
> familiar with it.
>
> 4. It is connected to a LAN.
>
> Would anyone care to guess what on earth is resetting my system clock,
> or even where I would begin to look? Thanks in advance,
>
> Chris
When I first installed this machine, I had a similar problem. My machine
was built and tested on the West Coast, so the time was 3 hours off. I
run my hardware clock on UTC, so I am 5 hours slow when on "God's" time
and 4 hours slow when on EDST. I could set the HW clock all I wanted, and
it still screwed up.
The problem is that /etc/adjtime contains data that helps the system
adjust for the rate of going of your HW clock, and this works only if the
error is relatively small. If you get a 3 hour error, it may evenutally
synch up, but it is faster if you just delete that file. It will be
recreated next time you boot the system or change run levels, IIRC. On my
Red Hat system, /etc/rc.d/rc.sysinit is some code that tries to re-adjust
the clock (/sbin/hwclock) and if /etc/adjtime is too far off, it will
screw up your hw clock in attempt to fix things.
If you are running Linux only, I suggest running your hardware clock on
UTC. If you are dual booting with the proprietary binary-only
operating-system, you will be stuck needing to use what they need.
--
.~. Jean-David Beyer Registered Linux User 85642.
/V\ Registered Machine 73926.
/( )\ Shrewsbury, New Jersey
^^-^^ 9:05pm up 20 days, 4:34, 1 user, load average: 0.99, 0.97, 0.91
------------------------------
** 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
******************************
