Linux-Advocacy Digest #500, Volume #32 Mon, 26 Feb 01 14:13:06 EST
Contents:
Time for a Windows reinstall! (Aaron Ginn)
Re: RTFM at M$ (Bob Hauck)
Re: RTFM at M$ (Bob Hauck)
Re: Something Seemingly Simple. (Bob Hauck)
Re: Time for a Windows reinstall! (Donn Miller)
Re: why open source software is better (dubin david scott)
Re: Check out this Windows bug (The Ghost In The Machine)
----------------------------------------------------------------------------
From: Aaron Ginn <[EMAIL PROTECTED]>
Subject: Time for a Windows reinstall!
Date: 26 Feb 2001 11:05:23 -0700
Well, I've finally run into a problem with Windows that I can't
solve. Yesterday, I was copying a CD on Windows using Nero. During
the copy, I saved the ripped image to disk and then copied to my new
CD. Nero was supposed to delete the CD image after the write but it
didn't. Yesterday, my C: drive filled up because the CD image was not
deleted. Now I'm having all kinds of problems with Windows, the most
annoying of which is that my scanning software does not see my scanner
anymore. I used the scanner before the CD write and it worked fine.
Now I can't get the software to even find it. I tried reinstalling
the drivers, changing the BIOS settings on the parallel port, and
changing the cable connection to every possible configuration.
Nothing worked.
I've had Windows on this computer for almost two years without a
reinstall, so I'm actually surprised it has lasted this long.
Performance has reached an absolute low: cursor pauses of about 5-10
seconds, 3-4 blue screens a day, the inability to run more than one
program at a time without completely locking up the computer.
If anyone has any suggestions, I'd be happy to hear them. My last
hope is to run Scandisk and Disk Defragment in hopes that the problem
will fix itself somehow, but I'm not confident.
I actually don't hate Microsoft. I just think it's ridiculous that a
software company with so much cash and so many talented programmers
could churn out such shoddy products. Even though Linux has fewer
apps available, at least it's built on a solid foundation. Windows is
the house built on sand.
Aaron
--
Aaron J. Ginn Phone: 480-814-4463
Motorola SemiCustom Solutions Pager: 877-586-2318
1300 N. Alma School Rd. Fax : 480-814-4463
Chandler, AZ 85226 M/D CH260 mailto:[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: alt.destroy.microsoft
Subject: Re: RTFM at M$
Reply-To: hauck[at]codem{dot}com
Date: Mon, 26 Feb 2001 18:40:59 GMT
On Mon, 26 Feb 2001 04:14:28 GMT, T. Max Devlin <[EMAIL PROTECTED]>
wrote:
>Said Erik Funkenbusch in alt.destroy.microsoft on Sun, 25 Feb 2001
>> I first heard of Smurf attacks almost 3 years ago.
> Such an "attack" is only possible if you have *no* firewalling.
> Simply ensuring that a single router won't pass packets with obviously
> bogus source addresses is 100% protection.
You're missing something there Max. The source addresses aren't bogus,
they are just not the ones from which the initial ping actually came.
You're right that any competent network admin will not allow such forged
packets to leave his network. But there are a lot of incompetent admins
out there whose network can be used to launch this attack, so a
defense-in-depth strategy makes a lot of sense.
You've got to be careful arguing with Erik. He does have a few clues.
--
-| Bob Hauck
-| Codem Systems, Inc.
-| http://www.codem.com/
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: alt.destroy.microsoft
Subject: Re: RTFM at M$
Reply-To: hauck[at]codem{dot}com
Date: Mon, 26 Feb 2001 18:36:06 GMT
On Mon, 26 Feb 2001 00:36:17 GMT, T. Max Devlin <[EMAIL PROTECTED]>
wrote:
>Said Bob Hauck in alt.destroy.microsoft on Sun, 25 Feb 2001 19:43:44
>>There _have_ been attacks based on pinging broadcast addresses using
>>packets with forged source addresses. Blocking ICMP to the broadcast
>>address is a sensible thing to do and doesn't break any functionality.
> Actually, there is a great deal of value in supporting ping to
> 'broadcast addresses' (it doesn't really broadcast, by the way)
What happens is that all the machines on the subnet answer. This is the
basis of the "smurf attack". It is used to DoS someone by pinging a
third party's broadcast address. You fake the source address on your
ping so that the responses go to your victim. Now you can ping-flood
someone from a dialup line without revealing your IP address or using
any of your own bandwidth.
> "Blocking" such a thing would be a monumental waste of time, as is all
> of this paranoid mucking around with IP.
CERT and Cisco both recommend that you filter ICMP to broadcast
addresses at your border. The recommend this because of the smurf
problem.
--
-| Bob Hauck
-| Codem Systems, Inc.
-| http://www.codem.com/
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: comp.lang.c
Subject: Re: Something Seemingly Simple.
Reply-To: hauck[at]codem{dot}com
Date: Mon, 26 Feb 2001 18:44:38 GMT
On Sun, 25 Feb 2001 20:25:29 +0000, Edward Rosten <[EMAIL PROTECTED]> wrote:
>No, that would be horribly slow, calculating PI every time you needed it.
I know that, but the question was whether there was an official place
where PI is defined in the library, not about a fast way to get PI.
<grin>
--
-| Bob Hauck
-| Codem Systems, Inc.
-| http://www.codem.com/
------------------------------
Date: Mon, 26 Feb 2001 14:02:27 -0500
From: Donn Miller <[EMAIL PROTECTED]>
Subject: Re: Time for a Windows reinstall!
Aaron Ginn wrote:
> could churn out such shoddy products. Even though Linux has fewer
> apps available, at least it's built on a solid foundation. Windows is
> the house built on sand.
The house that David Cutler built. As I understand it, Cutler had
something to do with DOS as well. I know Cutler designed rsx-11, and
CP/M is derived from it. DOS is derived from CP/M, and hence DOS has an
RSX-11 ancestry. You can actually see some VMS-isms in DOS, such as the
"dir" command, and VMS-style switches (/w /p, etc.) Also, I believe VMS
has special device names like LPT1:, COM1:, and CON:. This is assuming
that Cutler carried some of the RSX-11 design into VMS, of course.
Also, note the horrible 3-char extension that VMS has: VMS has
PROGRAM.EXE, etc., same as WinDOS.
Then, DOS was given some unix-like commands later, as I understand it.
The above is from what limited I know about VMS, DOS, RSX-11, so I'm
sure I'm missing something. So, both Windows 95 and NT are both
Cutler-esque (since 95 is essentially DOS). Aside: I wonder what
Cutler thinks about NT and 95? Hmmmm...
If I'm wrong, EF will appear in his Superman costume (with the initals
EF instead of "S") and save the day. He's quite a guy, that EF.
====== Posted via Newsfeeds.Com, Uncensored Usenet News ======
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
======= Over 80,000 Newsgroups = 16 Different Servers! ======
------------------------------
From: dubin david scott <[EMAIL PROTECTED]>
Subject: Re: why open source software is better
Crossposted-To: gnu.misc.discuss,comp.os.ms-windows.advocacy,misc.int-property
Date: Mon, 26 Feb 2001 18:57:00 GMT
In gnu.misc.discuss Ernest Schaal <[EMAIL PROTECTED]> wrote:
[...]
> While I would agree that there are advantages to open source
> software if you are a programmer, those advantages don't apply for
> the computer user.
Mr. Schaal, I believe that open source software affords direct
benefits to users, not just programmers, and that your example of
desktop publishing offers a good illustration for my points.
> When I started using computers (1962), computer programmers and
> computer users were synonymous because we had to write the programs
> we ran. For instance, I started out with Fortran, and later moved to
> BASIC, Pascal, etc. Nowadays, it doesn't make sense to write all
> your programs, or to spend time modifying them.
Proprietary software has traditionally offered you two choices: write
an application entirely on your own, or pay to use a complete
application that has been written for you. With free and open source
source software, it's possible to build on components and resources
written by others. This is a different perspective on progress since
1962: a person with modest programming capability can accomplish so
much more than they used to. One can obtain, for example, the major
components of a desktop publishing system (advanced editor,
high-quality typesetting engine, print previewer, SGML/XML parser,
revision control, etc.) all under the terms of liberal, open source
licenses, and link them together with very simple scripts or utility
programs.
You may well ask why users should do even a little programming when
monolithic, integrated packages (like MS Word) are available. Indeed,
what about users who can't program at all? Before answering, I'd like
to point out that even proprietary software publishers like Microsoft
are moving to component-based development, now that it's possible to
distribute services without releasing source code.
You wrote:
> For instance, I use Microsoft Word a lot. When I started with the earlier
> versions of Word (late 1080s), I also needed PageMaker in order to do
> newsletters and brochures. Since that time, Word has improved to the point
> that I do significant desktop publishing with it.
I'm a member of an international scholarly society, and part of my
participation involves helping to distribute their newsletter, which
is prepared using an extremely popular and ubiquitous commercial word
processing program that I won't name. The nature of the publication,
its contributors, audience, and distribution methods subject the
document to stresses that your publishing projects may not face (or
maybe they do -- I can only speak from my own experience).
Although the document is written in English, the names of many persons
and places are written in non-Latin alphabets. The same goes for
titles of books and papers that are referenced or cited. The
publication's audience accesses it over the web using many different
kinds of hardware and software, customized for a variety of locales.
Converting the document between different file formats is important
if everyone in the society is going to be able to read the newsletter.
For all the advanced features this word processing software offers,
the robustness we need for our newsletter seems difficult to achieve:
some design, editorial, or content expression inevitably breaks when
the file is shared or converted. I think reasonable people can
disagree on the extent to which this is the fault of the software or
the users. But wherever you come down on that issue, I think
robustness under all kinds of stress is one of the main advantages of
open source software over proprietary software (for reasons I'll
describe in further detail below). I think users benefit from that
robustness in a very direct way.
The SGML-based publishing system that I use for my own projects gives me
none of the problems I cited. Extended character sets and graceful format
translation are handled with little hassle. I attribute this to several
factors:
1) The open source components from which I assembled the system
were written for general-purpose use, not just the needs of one
application.
2) The maintainers of those components enjoy a more direct feedback
relationship with their user base than the authors of
proprietary software. And since the user base itself is more
sophisticated, the feedback they can offer is more constructive.
3) The components have to have good support for standardized
interchange formats and protocols, since open source developers
obviously can't rely on proprietary protocols.
4) Releasing your source code invites people to stress it in ways
you wouldn't think anyone would want to. As a result bugs get
discovered and corrected quickly.
5) As a user of the system, I have to be a *little* more educated
about the underlying representation of my documents, even if I'm
not a software engineer. I have to understand the relationship
between the character set standard I'm using and its encoding
as a stream of bytes. I have to understand how the logical and
structural elements of the document are mapped to formatting and
design elements. This is not to say my software isn't as "fully-
featured" as MS Word. But the sophistication is invested in
more flexible editing, more advanced typography, and better
project management features, rather than hiding details behind
the illusion of seamless integration. I think that's
representative of different priorities in the open source
community: user-empowering and user-educating, instead of merely
user-friendly.
> I have seen Word grow in size and features, from a simple program
> that allowed be to do simple word processing to a complex program
> that allows be to do so much more. While I don't like complexity for
> its own sake, I do enjoy the ability to do my work more productively
> by using full-featured software.
Sure. And I'll concede that proprietary software publishers have had
great success in writing and marketing applications for productive
non-programmers. In the open source community, software for capable
programmers comes first, software for casual programmers comes later,
and software for non-programmers comes last of all. But the vehicles
by which open source programmers arrive at that point bring their own
built-in advantages.
Dave Dubin
GSLIS, UIUC
------------------------------
From: [EMAIL PROTECTED] (The Ghost In The Machine)
Subject: Re: Check out this Windows bug
Date: Mon, 26 Feb 2001 19:04:46 GMT
In comp.os.linux.advocacy, Aaron Kulkis
<[EMAIL PROTECTED]>
wrote
on Wed, 21 Feb 2001 17:24:21 -0500
<[EMAIL PROTECTED]>:
>
>
>The Ghost In The Machine wrote:
>>
>> In comp.os.linux.advocacy, Aaron Kulkis
>> <[EMAIL PROTECTED]>
>> wrote
>> on Tue, 20 Feb 2001 17:12:54 -0500
>> <[EMAIL PROTECTED]>:
>> >
>> >
>> >Donn Miller wrote:
>> >>
>> >> Aaron Kulkis <[EMAIL PROTECTED]> wrote:
>> >>
>> >> > He was attempting to prepare and move files.
>> >> > What's to question?
>> >>
>> >> I can't believe how bad FAT32 is! Damn, is there a worse filesystem out
>> >> there:
>> >
>> >FAT16
>> >FAT12
>> >
>> >
>> >> severe fragmentation, no symlinks, long file names is really a hack,
>> >> reliability is horrid. OTOH, I will give FAT credit for one thing. Isn't it
>> >> the best filesystem for very small filesystems, like say, for a 1.44 MB
>> >> floppy, for example? You know, if someone in 1991 (when I first started
>> >
>> >ext2fs on a floppy seems to hold more.
>>
>> A small test case:
>>
>> /h5/ewill/dumb/flop.ext2
>> 1412 13 1327 1%
>>
>> /h5/ewill/dumb/mount.ext2
>>
>> /h5/ewill/dumb/flop.fat
>> 1423 0 1423 0%
>> /h5/ewill/dumb/mount.fat
>>
>> might be construed either way. (Don't pay too much attention to the '1327';
>> ext2 has a reserve factor that allows the superuser to write blocks when
>> ordinary users run out. The default is 5%. Also, I've folded the
>> lines to fit; otherwise, they'd run off the end of the 80-char
>> boundary. The mounts are using the loop device.)
>>
>> Without lost+found, it actually would hold more, and for a floppy,
>> lost+found -- which is the same size regardless of disk size; there
>> doesn't seem to be an elegant method to override it -- is somewhat
>> pointless.
>>
>> With lost+found, of course, the FAT floppy has 11 more blocks.
>>
>> Of course, even without lost+found, the number of blocks is only 2.
>> The cluster sizes also look the same.
>>
>> The only issue that ext2 has is that user and group IDs are supported.
>> However, this is slightly pointless on a floppy (root can read
>> anything). :-)
>>
>> There might be an issue with directory entries; on vfat (fat+Win95),
>> the short entries are the same, but the long entries take up twice
>> as much space per filename because it uses Unicode characters.
>> This may be partially offset by bigger inodes for ext2, but I'd have
>> to look up the details. I do find the vfat storage method hackish
>> and wasteful, however.
>
>
>What was your ratio of (512 or 1k) blocks to inodes?
I didn't set it explicitly; I can tell you the following:
$ /sbin/mke2fs ./flop.ext2
mke2fs 1.18, 11-Nov-1999 for EXT2 FS 0.5b, 95/08/09
./flop.ext2 is not a block special device.
Proceed anyway? (y,n) y
Filesystem label=
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
184 inodes, 1440 blocks
72 blocks (5.00%) reserved for the super user
First data block=1
1 block group
8192 blocks per group, 8192 fragments per group
184 inodes per group
Writing inode tables: done
Writing superblocks and filesystem accounting information: done
I can also tell you that the root area of the DOS floppy is either
112 or 224 entries. A small PERL program created 224 files before
abending. (fat has no limit on total inodes on subdirectories, or
total inodes, period, beyond the number of clusters on the physical
drive; however, the root directory, for inscrutable reasons, is a
contiguous area on the floppy/hard drive, with a fixed number of
entries. Evidently someone thought it a good idea at the time...)
>
>conduct the test again using a reasonable number of inodes
>[1000 or so i-nodes for a floppy disk seems rather...excessive...to me]
A similar run created 172 files on the ext2 floppy. (I'm not sure
where the other inodes went; I should have 182. Oh well.)
After creation, 1311 blocks were still "available" on the fat
floppy, whereas 1152 blocks were "available" on the ext2 floppy.
Apparently fat is using a blocksize of 512 bytes.
(Each file was about 11 bytes or so long.)
Note that this was a straight fat. If I mount using -t vfat, the
results change if the Unicode variant is required (###.JNK doesn't
require the Unicode additional entry/ies, but ###.sillyjunk does).
I can now only create 112 files, and the blocks available goes up
to 1367, on the vfat floppy -- mostly because it ran out of direntries first.
It gets even worse, the longer the filenames become.
If I do an alteration to generate filenames of the form
PrefixForThis_###.moresillyjunk
where ### is where the number goes, I get 56 files on the vfat
floppy, whereas the number of files on ext2 doesn't change.
Obviously, I can get arbitrarily silly with this, up to
the maximum name length ext2 can support.
Note that this does not mean fat is a better format; it's not.
It is more efficient spacewise for data storage (ONLY because
fat floppies use 512 byte blocks!), but not for filename/inode storage --
and heaven help the poor user who gets his fat chains cross-linked.
(In the case of ext2, a user might get a live block mismarked as free
and that block subsequently gets allocated and blasted -- which toasts
the file [*] contained in the block. However, that's one block out of
one file, whereas fat crosschaining destroys two files and an indefinite
number of blocks.)
It is also horrid for name storage -- a quick peek at the actual
"floppy" reveals that names are stored in the form:
0023100 001 P \0 r \0 e \0 f \0 i \0 017 \0 - x \0
0023120 F \0 o \0 r \0 T \0 h \0 \0 \0 i \0 s \0
0023140 P R E F I X ~ 1 M O R 030 \0 005 M
0023160 Z * Z * \0 \0 005 M Z * 302 001 \t \0 \0 \0
0023200 C u \0 n \0 k \0 \0 \0 377 377 017 \0 M 377 377
0023220 377 377 377 377 377 377 377 377 377 377 \0 \0 377 377 377 377
0023240 002 _ \0 1 \0 . \0 m \0 o \0 017 \0 M r \0
0023260 e \0 s \0 i \0 l \0 l \0 \0 \0 y \0 j \0
apparently so that Unicode characters can be nicely handled --
if one can call this nice.
[rest snipped]
[*] The file could also be a directory, which means that inodes underneath
that directory may become inaccessible. However, that's true regardless
of filesystem.
--
[EMAIL PROTECTED] -- insert random misquote here
EAC code #191 21d:03h:53m actually running Linux.
The US gov't spends about $54,000/second. I wish I could.
------------------------------
** 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 by posting to comp.os.linux.advocacy.
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
