Linux-Development-Sys Digest #350, Volume #6 Sat, 30 Jan 99 15:14:06 EST
Contents:
Re: Why I'm dumping Linux, going back to Windblows (Dave Hulsopple)
Re: disheartened gnome developer (jedi)
Re: Why I'm dumping Linux, going back to Windblows (Stan Vermaelen)
Re: How to flush the file cache ? (Jeff McWilliams)
Sockets (Robert Norman)
Re: Why I'm dumping Linux, going back to Windblows ([EMAIL PROTECTED])
Re: COM for Linux? ("Aaron Perrin")
kernel problem (Yuhe Liu)
Re: trying to write a module which uses parallel port, but inb( port ) (Olof
Wolgast)
Linux and nfs3 (Cameron Schaus)
Re: Printing to HP DeskWriter 560C (Matt Kressel)
Re: glibc-performance (Andreas Jaeger)
----------------------------------------------------------------------------
From: Dave Hulsopple <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Why I'm dumping Linux, going back to Windblows
Date: Fri, 29 Jan 1999 11:17:18 -0600
"Paul D. Smith" wrote:
>
> %% [EMAIL PROTECTED] writes:
>
> b> In article <[EMAIL PROTECTED]>, doug says...
>
> >> Linux does not have to be for everyone, and perhaps should
> >> not. Doug
>
> b> Linux can be made easier to millions more with very small effort.
> b> Such as what people are talking about here: adding examples in man
> b> pages.
>
> b> We can do small things that will have large effects to make Linux
> b> easier.
>
> b> offcourse, this all requires Linux people have more open mind about
> b> this, and not be too stubburn and want to close the doors to new
> b> users too quickly just becuase someone is walking in slower than
> b> they did.
>
> Look, no one is saying that Linux couldn't be simpler to use, or that
> beginners don't need examples of sometimes-arcane syntax in use, etc.
> That is all true and we can nod sagely together about it.
>
> The _question_ is, does this beginner-level documentation belong in the
> man pages? I think an excellent argument can be made that no, it does
> not.
>
> Man pages are like reference manuals: they describe the thing in
> technical terms and provide all the nitty details needed to use it: most
> of the time far more than the poor newbie user wanted to know.
>
> What's needed is _separate_ documentation which provides beginner-level
> help, but is as easily accessible as man pages. That means you should
> be able to say "help find" or something and get a find _tutorial_: what
> it does and how to use it on a basic level with examples of
> commonly-performed operations. Then the user, so fortified, could
> proceed to the man page for find in all it's glory, if they didn't see
> what they wanted in the beginners' help text.
>
> If you try to squash user-level info into reference man pages, you'll
> just piss off both the beginning folks who will still be offput by the
> grody details, _and_ the more advanced folks who just want to know what
> arguments select() takes and what error code it will return if there is
> no data available and don't need no mushy introduction text.
>
> Luckily, I'm sure there are already a number of good "intro to unix"
> documents around; someone just needs to combine them and create a "Unix
> for Beginners" package that RedHat, Debian, etc. can include with a
> standard distribution. They could point users to it both in the
> documentation and even in the install/setup information.
>
> So... go for it, y'all!! :)
>
> Anyway, that's my $0.02.
>
> --
> -------------------------------------------------------------------------------
> Paul D. Smith <[EMAIL PROTECTED]> Network Management Development
> "Please remain calm...I may be mad, but I am a professional." --Mad Scientist
> -------------------------------------------------------------------------------
> These are my opinions---Nortel Networks takes no responsibility for them.
==========================================================================
Howdy all!
Having followed this thread for a while now, it occurs to me that
what you want, I may already have.
As part of a training course I did recently, I put together a longish
document that takes the newbie from "soup to nuts" and leaves them with
a good grasp of what unix is in general, what Linux is in specific, and
takes the student through the basics and beginnings of both, including
the "core 15" command they will use on a regular basis.
The actual course takes about three days to administer, but the docs
are much smaller, since I do most of my teaching from memory.
This document could probably use some work (of course), but I would
be willing to make it available to any and all who might want it.
Including it into any distribution of Linux would be dead bang simple,
since it only ascii text anyway, but some smart programmer type could
probably find a way to have it pop up on demand, so that the newbie
could use it as a "primer", the move on to the man pages, once they
become comfortable with the system.
I must say up front that I ganked part it from various sites around
the internet (attributes and URL's available upon request), but much of
it is in my own hand, so plagerism isn't an issue.
The course has been well received, and my current employer is
attempting to have me teach it to FE's that need that little extra boost
to supporting Unix.
Please let me know if I can help.
HTH
Dave Hulsopple
------------------------------
From: [EMAIL PROTECTED] (jedi)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.x
Subject: Re: disheartened gnome developer
Date: Fri, 29 Jan 1999 09:19:28 -0800
On 29 Jan 1999 10:03:28 GMT, Mike Willett LADS LDN X7563
<[EMAIL PROTECTED]> wrote:
>In article <[EMAIL PROTECTED]>,
> [EMAIL PROTECTED] (Marcin Krol) writes:
>>On Sun, 24 Jan 1999 14:03:47 GMT, [EMAIL PROTECTED]
>>wrote:
>><snip>
>>>>This may be true ideally, but I certainly never heard any evidence
>>>>that it works in practice. Ask any recent emmigrant from the Union of
>>>>Soviet Socialist Republics how efficiently their economic system was
>>>>working when they left.
>>
>>>About the only way in which the USSR was socialist is in name.
>>
>>No, it was not.
Even the Soviet Leaders only ever claimed to be BUILDING
communism and never claimed to have achieved it. Even by
their own explicit admission, their government/nation was
not in it's desired end state.
>>
>
>From what I read of Socialism - USSR was NOT a true socialist state
>What features of USSR do you think it is a true socialist model.
>
The USSR was state capitalism brought about to help
further the eventual development of a truely communist
state.
--
Herding Humans ~ Herding Cats
Neither will do a thing unless they really want to, or |||
is coerced to the point where it will scratch your eyes out / | \
as soon as your grip slips.
In search of sane PPP docs? Try http://penguin.lvcm.com
------------------------------
From: [EMAIL PROTECTED] (Stan Vermaelen)
Crossposted-To: alt.os.linux,comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Why I'm dumping Linux, going back to Windblows
Date: Sat, 30 Jan 1999 13:25:30 GMT
>
May I contribute a couple of Eurocents to this discussion?
I think man pages are ok to document command lines (and not GUI-based
programs). Man pages cannot teach concepts. Man pages provide a
consistent mechanism to deliver the information needed by various
types of users at the moment they need it.
In my experience, man page writers often not sufficiently take their
audiences into account. For instance, a program developer tends to
ramble on about all the clever things his piece can do, things in
which the user may not always be interested ;).
It also surprises me that on many man pages I read messages like:
"This man page is out of date, read the Texinfo". Which leaves me in
the cold, since I don't have such a bug in my box (I don't even have
Emacs )->), and leaves me with yet another doc format to deal with
besides man, html, sgml, info and plain ascii. And leaves me without
the use of commands like 'apropos' and 'man -k'.
Perhaps information delivery is some sort of black art; I'm not sure
if it will ever be possible to write documentation to suit everybody's
needs. Increasing the volune doesn't help, quite on the contrary: if
for any user, the reading matter becomes too voluminous and too
detailed, it stops being information and becomes unintelligable
gibberish. Meaning that anyone who says that Linux (Unix) is well
documented because there exist thousands of man pages and hundreds of
books, and that newbees are stupid because they refuse to RTFM, is
talking nonsense.
I think that man pages will do just fine if:
1. Developers take as much care in designing their docs as they do in
designing their programs. Documentation is after all part of the
finished product.
2. Man page writers take their various audiences into account. Like:
casual users who need to know the purpose and use of the command, and
lots of practical examples; professional sysadmins who need formal and
detailed knowledge on the use of the program; programmers, who may
want to know the basics; and perhaps there exist other audiences. I
believe that's why there exist levels in the man page structure.
3. Man page writers realize the limitations of the medium. It's not
possible to completely describe all the ins and outs of a complex
application in a few man pages (we have O'Reilly for that...).
Personally, I like the consistency of the man page mechanism. If
properly designed, man pages are a very nice medium to deliver the
information that a command line user needs, at the moment (s)he needs
it. But man pages can not teach concepts; for this I think courses,
books and electronic media (HTML, etc) are better.
--
Stan
Speaking as a sometimes somewhat mystified Linux user.
------------------------------
From: [EMAIL PROTECTED] (Jeff McWilliams)
Subject: Re: How to flush the file cache ?
Date: 30 Jan 1999 13:31:55 GMT
In article <[EMAIL PROTECTED]>, Renaud Lottiaux wrote:
>Is anybody know how to flush the file cache
>quickly and easily ?
Try issueing the sync command. that's what I do right before I want to
unmount a zip disk and right after I've just copied a bunch of files to
it. It forces the system to complete all pending writes... at least
that's how it works for me. See the man page for sync
Jeff
--
Jeff McWilliams
"The mystery boy has gone to war, in the fields of death you'll find him."
------------------------------
From: Robert Norman <[EMAIL PROTECTED]>
Subject: Sockets
Date: Sat, 30 Jan 1999 14:02:35 +0000
Hi,
I'm writing a Java Client/Server program. The server resides on a Linux
box and allows clients to connect to it. Each client connects using a
socket.
The problem is that Linux seems to be limiting us to 240 socket
conenctions. Is there a way around this?
Any help would be greatly appreciated.
Cheers,
Rob.
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: alt.os.linux,comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Why I'm dumping Linux, going back to Windblows
Date: 29 Jan 1999 09:52:05 -0800
In article <[EMAIL PROTECTED]>, Chris says...
>
>Go buy a book that has examples in it and use the man pages as a
>reference. I think that's all they are intended for.
This is a silly argument thrown with no thinking behind it.
"go buy a book with examples"
It seems that people has lost all common sense.
Now, tell us, which books should I buy that will give me an example of how
to use each Unix command?
How much will these books cost?
I think people are mixing two main issues here, the first is that one
should offcourse understand concepts and ideas about the system, and
this is the long term goal, and something that happen with time and
by reading books.
The second is the issue of how to help people figure how to do things in
the short term, so that not every little bit of difficulty becomes an
all time consuming task.
Some of the good books to help someone learn about Unix are:
Unix power tools.
Unix secrets
essential system adminstration
Linux the complete refererence
Unix for the impatient
Linux man by redhat
Now, man pages with examples help one in the second case above, while books
will help with the first case. It is also not the case where one will
have a book next to them when they need it. and it is not the case that
the book they will have will contain an example of how use the command
they need to use at the time. And it is also much faster to have the examples
in the man page where one can quickly find it.
Bill.
------------------------------
From: "Aaron Perrin" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: COM for Linux?
Date: Sat, 30 Jan 1999 12:29:24 -0400
Thanks for posting the FAQ URL.
I know I probably should have looked for FAQ first before posting,
but I assumed (incorrectly, of course) that most people using linux
wouldn't be interested in such a thing. Oops.
Basically I haven't looked at CORBA for awhile, and the last time I
checked many of the ORBs available were implementing features
that were beyond the current standard (2.0 maybe?). There was
some controversy about the maturity of CORBA and also the fact
that ORBs cost several thousand dollars, and I decided to use COM
instead. Anyway, I have some COM components that I'd like to utilize
on a new linux box I'm planning on buying, and I didn't really want to
rewrite or recompile for CORBA.
I'll admit that I haven't been keeping up-to-date on CORBA, so I'll
check it out and reconsider, but either way, thanks for the info.
Aaron
[EMAIL PROTECTED]
------------------------------
From: Yuhe Liu <[EMAIL PROTECTED]>
Subject: kernel problem
Date: Fri, 29 Jan 1999 12:58:04 -0600
Hi there,
I got a problem today on my RedHat 5.2 box. The system had been running
for 40 days. Today, just a sudden, the system seemed frozen. The load
chart rushed up to maximum. I switched from X to console. There were
contiuous printout of a same statement on the console:
Problem: block on freelist at 00c36010 isn't free.
A few minutes later, the system came back to normal. I checked the
system
log. There were similar statements found in file message:
Jan 29 11:13:45 ocean kernel: Problem: block on freelist at 00c36010
isn't free.
Jan 29 11:13:54 ocean last message repeated 9450 times
Jan 29 11:13:54 ocean kernel: k on freelist at 00c36010 isn't free.
Jan 29 11:13:54 ocean kernel: Problem: block on freelist at 00c36010
isn't free.
Jan 29 11:13:54 ocean last message repeated 74 times
...
Jan 29 11:16:12 ocean kernel: Problem: block on freelist at 00c36010
isn't free.
Jan 29 11:16:12 ocean last message repeated 74 times
Jan 29 11:16:12 ocean kernel: Probem: block on freelist at 00c36010
isn't free.
Jan 29 11:16:12 ocean kernel: Problem: block on freelist at 00c36010
isn't free.
Jan 29 11:16:12 ocean last message repeated 73 times
Every thing looked normal, nothing seemed happened except the log
records.
Was it related to the filesystem? a kernel bug? system bug? application
software bug? or did it mean a security compromise of the system?
Hopefully some one could explain it to me. Very appreciate.
Thanks.
Yuhe Liu
------------------------------
From: Olof Wolgast <[EMAIL PROTECTED]>
Subject: Re: trying to write a module which uses parallel port, but inb( port )
Date: Fri, 29 Jan 1999 21:33:46 +0100
I've done more research. It seems like the value written to 0x378 affects
port 0x379. For example, if 3 is written to 0x378, 0x379 becomes 110. Can
anyone explain this? It really goes beyond my knowledge
Olof Wolgast wrote:
> I got my hands on a module which is supposed to create a link to your
> texas-calculator using the parallel port. It didn't work, so I sat down
> and went searching in the code. I'm not at all experienced in kernel
> programming, but I tracked it down to this: inb( port ) always returned
> 74. It doesnt matter what was written in the other end or what is
> written from the computer, the return value from inb is always 74. I
> know that the hardware is correct, because it works under dos, and the
> shematic for the cable is very easy to check with an ohm-meter. I
> haven't any support for lp in the kernel, so that shouldn't affect. What
> can be the problem? The code is about 1.5 years old, does that old code
> work with recent( 2.0.36 ) kernels? I've heard the reverse.
>
> Besides, is there a good way to print out variables from the kernel?
> Like you can in a normal program with printf
>
> Grateful for help
>
> //Olof Wolgast
------------------------------
From: [EMAIL PROTECTED] (Cameron Schaus)
Subject: Linux and nfs3
Date: Fri, 29 Jan 1999 20:42:15 GMT
I have a linux machine that mounts nfs filesystems from an aix4.3.2
machine, which is using nfs3.
Reads from the mounted filesystems are somewhat fast, but the dismal
speed of writes is something I want to fix. Both machines are on a
100Mb/s ethernet, yet writes are typically around 70Kb/second
(painfully slow). I have tried all sorts of options when mounting the
filesystems, but none have helped.
I tried using the 2.2.0-pre4 kernel, but that still didn't help.
I have also been reading in the newsgroups about this problem, and
there seems to be an acknowledgment of the problem, but there are no
solutions (either current or pending) to be found.
What is the state of the development in this area, a linux client
mounting from an nfs3 server? Are there any solutions on the horizon?
I would really like to get this fixed since this linux machine can be
really fast without slow nfs holding it back.
Thanks,
Cameron
------------------------------
Crossposted-To: comp.os.linux.networking
From: Matt Kressel <[EMAIL PROTECTED]>
Subject: Re: Printing to HP DeskWriter 560C
Date: Fri, 29 Jan 1999 21:08:55 GMT
Scott Nielson wrote:
>
> I have been using the cdj550 device in ghostscript 4.30 to print to a HP
> DeskWriter 560C. It works, but before the job and after each page I am
> getting an extra sheet with one line on it. The line will say something
> like: "gnu ghostscript 4.30" or "Page 1" or "Last Page".
>
> Am I using the right ghostscript device? Or, is there a better one?
You are using the right one, but perhaps have the wrong options. Look
into apsfilter. It sets up many HP 5xx family printers fairly easily.
-Matt
--
Matthew O. Kressel | INTERNET: [EMAIL PROTECTED]
+--------- Northrop Grumman Corporation, Bethpage, NY ---------+
+--------- TEL: (516) 346-9101 FAX: (516) 346-9740 ------------+
------------------------------
From: Andreas Jaeger <[EMAIL PROTECTED]>
Subject: Re: glibc-performance
Date: 30 Jan 1999 19:14:29 +0100
>>>>> Horst Piening writes:
> because of the new kernel i started updating my system to glibc2.06pre7:
> got new binutils, gettext, texinfo ..
> installed gcc 2.8.1 and made the glibc ( with --disable-profile ! )
--enable-profile generates an extra set of libraries and doesn't have
any impact on the normal libs.
> gcc -v:
> Reading specs from /usr/lib/gcc-lib/i586-pc-linux-gnulibc1/2.8.1/specs
> gcc version 2.8.1
> ldd ./rayn
> libmm.so => ./libmm.so
> libobj.so => ./libobj.so
> libea.so => ./libea.so
> libtree.so => ./libtree.so
> libtex.so => ./libtex.so
> libtest.so => ./libtest.so
> liball.so => ./liball.so
> libm.so.6 => /usr/i586-linuxglibc2/lib/libm.so.6
> libc.so.6 => /usr/i586-linuxglibc2/lib/libc.so.6
> ld-linux.so.2 => /lib/ld-linux.so.2
> ldd ./rayn
> libmm.so => ./libmm.so
> libobj.so => ./libobj.so
> libea.so => ./libea.so
> libtree.so => ./libtree.so
> libtex.so => ./libtex.so
> libtest.so => ./libtest.so
> liball.so => ./liball.so
> libm.so.5 => /lib/libm.so.5.0.6
> libc.so.5 => /lib/libc.so.5.4.7
> ** with gcc 2.8.1 , glibc2.06pre7 : 181.10 user 0.19 system 3:13.16elapsed
>93%CPU
> ** with gcc 2.8.1 , libc.so.5 : 150.2 sec
And what's the exact timing with libc.so.5?
> when i ran my raytracer for time testing the new glibc stuff was
> 15 % slower than before.
> what did i overlook, i expectet a time increase ?
Raytracer? With which compiler options did you compile it? It might
be that some of the math functions are slower - but they're now
handling all special cases correct. If you could profile both
programs (you don't need profiling libs for that), I'd be interested
where the time is spent. Different compilers and compiler options for
building the two libs might also result in the difference.
Another point is that most of glibc2's functions are threadsafe. This
results especially in a slowdown when using fgets and friends. If you
use those functions, you might consider using the *unlocked variants
(see <stdio.h>).
Andreas
--
Andreas Jaeger [EMAIL PROTECTED] [EMAIL PROTECTED]
for pgp-key finger [EMAIL PROTECTED]
------------------------------
** 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.development.system) 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-Development-System Digest
******************************
