Linux-Development-Sys Digest #985, Volume #6 Fri, 23 Jul 99 02:14:14 EDT
Contents:
Re: HELP: how to measure hard disk access performance on Linux? (John McKown)
Re: New OS development (John Stevens)
Compiling Java programs on Red Hat Linux 6.0 (Robert S Laramee)
Re: Tracing system calls (David B Anderson)
Re: Linux Training (M. David Allen)
Re: when will Linux support > 2GB file size??? (Donovan Rebbechi)
Re: when will Linux support > 2GB file size??? ("Chad Mulligan")
Re: Linux Training (Daniel Forester)
libc (jauming)
How to create a module? ("fernando Ortega")
Socket connection to Serial I/O - How? (Don Scheidt)
Re: "Contract Outsourcing?!" ("Margaret Dailey")
Linux hardware debugging (Helmut Artmeier)
Re: Asynchronous IO networking Stack? (Philipp Thomas)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (John McKown)
Subject: Re: HELP: how to measure hard disk access performance on Linux?
Date: 22 Jul 1999 23:57:49 GMT
On Thu, 22 Jul 1999 15:38:30 +0200, Dorin-Ioan MARINCA <[EMAIL PROTECTED]> wrote:
] >Hi!
] >
] >I want to know if exists some utils that can be used to obtain
] >performances for hard disks on Linux. I'm interested also for networking
] >subsystem.
] >Or maybe better... where can I obtain an updated doc for 'proc'
] >filesystem infos? Because there are some fields which maybe can be used
] >to calculate disk throughput [Mbps]:
[snip]
] >Thanks for any little hint,
] >D.I.M.
I'm running RH 6.0, but I think you can do this on 5.2 as well. Run the
command
hdparm -t -T /dev/hda
Replace /dev/hda with whatever disk you want to time.
------------------------------
From: [EMAIL PROTECTED] (John Stevens)
Subject: Re: New OS development
Date: Fri, 23 Jul 1999 02:18:20 GMT
On Thu, 15 Jul 1999 13:29:51 +0000, Igor Zlatkovic <[EMAIL PROTECTED]> wrote:
>Oh boy... look at the number of flames I got by now. And they are getting more
>every hour. Why couldn't I keep my mouth shut? Why the hell didn't I take the
>blue pill?
'Cause. . . you just *HAD* to see what would happen? HAD TO!?
Or. . . not. I'm easy! ;->
>I did it. Im impressed, but I didn't see any inheritance there. I saw a function,
>much like other functions in the kernel tree. It operates on big structures that
>contain pointers to other structures and functions. OK, I must admit that I
>didn't give it more than 10 minutes, but I couldn't find any language-supported
>inheritance there.
The mistake you are making is in misunderstanding the other guys point of
view.
C may not have direct, built in support for OO programming, but the assertion
wasn't that C has direct support for OO programming: the assertion was,
in paraphrase, that OO is a paradigm. OO can be done in any sufficiently
expressive language.
As for "inheritance" in C: please see the Xt book and source code.
As an example:
typedef struct {
int x;
void (*show)(FILE *fp);
} Core;
typedef struct {
Core super;
float y;
void (*newFunc)(fd_set set);
}
Or, another way of looking at this is:
Both C++ and Objective C were orginally implemented as preprocessors that
produced C code that was compiled by a C compiler into an executable.
And all OO code is expressed as machine code, which is almost (not quite)
one-to-one mappable to assembly, right?
>Overloading has a lot to do with OO.
Nope.
But support for subclass responsibility almost *IS* OO (Ok, a little
exaggeration, there, but it is really, really basic, and really, really
important).
>Being able to call operations implemented for derived objects through
>the pointer to a base object is the thing. You have a
>superobject with own implementation of some operation.
Ah. I see. You didn't mean "Overloading", you meant sub-class
responsibility.
>Now I can add as many subobjects as I like, each with its own version of
>the same operation without having to change or recompile the code that
>actually calls the operation.
Err. . . no. You meant, I suspect, "class", not object. Although,
strictly speaking, in a purely theoretical discussion, the difference
is minor. If not totally moot. Objective C supports true Class Objects,
including Class methods, for instance.
>If now every call calls
>the right version of the operation, then
>I have what I think polymorphism is and yes, this is very necessary.
Probably, again, not what you meant. Polymorphism is different from
subclass responsibility.
>Templates could be something that has less to do with OO. Templates
>are merely a type-safety concept.
Well. . . no, not really. Templates are a way to overcome some of
C++'s weaknesses. C++ is so strongly typed that genericity is
difficult to do properly without templates in C++.
IOW, it is more difficult to build a generic container class
(A List) in C++ than it is in other languages. Without templates
you almost always end up with (A List of <Type>). Hence STL.
Templates are not really needed in languages that support abstract
typing (or, at least, that have much better support for it).
>What I talk about is how nice it is having the language and the compiler
>natively support the way you
>thought when you designed the program. This makes the translation of the idea
>into code rather direct. You obviously understood it as a statement that the
>implementation of an OO designed software cannot be done without such language.
>This is my mistake, sorry. It can, and it has been done, but it always involved
>extra work.
Yup.
John S.
------------------------------
From: Robert S Laramee <[EMAIL PROTECTED]>
Subject: Compiling Java programs on Red Hat Linux 6.0
Date: Thu, 22 Jul 1999 22:41:03 -0400
Ever since Red Hat Linux 6.0 has been installed on my machine, my Java
programs do not compile anymore. Does anyone know why this is? It
looks
like there's a new compiler (Kaffe). I don't know why it doesn't seem
to
compile my Java programs. I get lots of classes missing messages even
though my classpath appears to be set properly. Is there anyone that
can
help me out with this? Is there any documentation about this problem?
--
Robert S Laramee tel: (603) 862-7603
14 McDaniel Drive, #422 email: [EMAIL PROTECTED]
Durham, NH 03824 URL: http://www.cs.unh.edu/~rlaramee
------------------------------
From: [EMAIL PROTECTED] (David B Anderson)
Crossposted-To: comp.os.linux.development,comp.os.linux.misc
Subject: Re: Tracing system calls
Date: 23 Jul 1999 03:04:55 GMT
In article <[EMAIL PROTECTED]>,
Gary Momarison <[EMAIL PROTECTED]> wrote:
>Mitch <[EMAIL PROTECTED]> writes:
>
>> I'm currently working on a BSDi box, and have recently been introduced
>> to a program called 'ktrace'. This thing is a great utility. My
>> question is whether or not there is something like this for linux? I've
>> made a cursory pass through freshmeat, but didn't turn up anything.
>
>http://www.aa.net/~swear/pedia/debugging.html had this link:
>
>http://www.linuxgazette.com/issue01to08/lg_issue6.html#strace
>
>to an article about "strace" which is probably similar to ktrace.
Other names (system dependent) are:
truss
par
They all seemingly have roughly similar
basic functionality, whatever the name.
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (M. David Allen)
Crossposted-To:
comp.os.linux.misc,comp.os.linux.networking,comp.os.linux.hardware,comp.os.linux.development.apps
Subject: Re: Linux Training
Date: 23 Jul 1999 03:12:30 GMT
Reply-To: [EMAIL PROTECTED]
In article <7n7qgv$iko$[EMAIL PROTECTED]>,
"JamesH" <[EMAIL PROTECTED]> writes:
> Hello,
>
> Can anyone recommend a good Linux Training Organization?
> Has anyone used or heard anything about Linuxcare or Redhat training?
>
>
>
> Jim Holloway
> GTS, Inc.
> 800 888-9874 ext. 203
>
Redhat does a linux certification program, I've heard that it's quite good,
but I know for sure that it's long and OBSCENELY expensive. Maybe that's just
me and my broke ass though. If you're part of a big company that can pay it,
why not?
Haven't heard anything about linuxcare.
--
David Allen
http://opop.nols.com/
========================================
People are like onions -- you cut them up, and they make you cry.
------------------------------
From: [EMAIL PROTECTED] (Donovan Rebbechi)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: when will Linux support > 2GB file size???
Date: 22 Jul 1999 12:47:04 -0400
On 22 Jul 1999 08:14:42 -0400, Robert Krawitz wrote:
>[EMAIL PROTECTED] (Christopher B. Browne) writes:
>I don't, because there isn't a critical mass of 64-bit hardware and
>software, and there isn't the kind of push behind it that there is
>behind Linux in general.
What about the Merced ? There is definitely "push", but not momentum yet.
Given that it's intel who are "pushing", I'd expect to see momentum in
due course.
--
Donovan
------------------------------
From: "Chad Mulligan" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.advocacy
Subject: Re: when will Linux support > 2GB file size???
Date: Thu, 22 Jul 1999 20:02:46 -0700
This question is a bit off topic but it just occurred to me that: NTFS will
support files of these sizes, and the current Linux kernels have support for NT.
Can they read and write to these files if an NTFS file system is used? I'm sure
that the whole file couldn't be cashed but a streaming handler should be able to
parse them, shouldn't it?
Christopher B. Browne wrote in message ...
>On 22 Jul 1999 08:14:42 -0400, Robert Krawitz <[EMAIL PROTECTED]> posted:
>>[EMAIL PROTECTED] (Christopher B. Browne) writes:
>>
>>> - It permits those that *need* big files to choose the appropriate
>>> hardware platform.
>>
>>Folks: in the real world, people don't have that choice! The choice
>>of hardware platform is dictated first by availability of applications
>>for it.
>>
>>People don't write and run applications (in the business arena, at any
>>rate) in a vacuum. They need infrastructure support: databases and
>>other middleware. Linux on Alpha or UltraSPARC is not an option for
>>these folks!
>
>I'm sorry, but there's a tough situation here, with no good "in between."
>
>a) Force those that need to have big files to choose their platform
>carefully, or
>
>b) Rewrite Linux's memory management system, with the associated period
>when it is resultantly desparately broken.
>
>Those are your options. Pick whichever one you dislike least. 64
>bits, or an OS that *doesn't work* for a year or so.
>
>>> - It encourages the deployment of 64 bit systems in those cases where
>>> 64 bits is in fact likely to provide benefit.
>>
>>From a business perspective, the Alpha looks like a dead end (Compaq's
>>hardly making an aggressive push), IA-64 doesn't exist, and SPARC V9
>>is expensive and already has Solaris.
>
>If your application is so important, then you're liable to need to
>choose from these anyways, whether you like it or not.
>
>>Notice that one consequence of all of this is that the business world
>>finds it very hard to make revolutionary changes. The only reason
>>Linux managed to gain traction here at all is that it is only an
>>evolutionary change from existing Unix platforms, but with some
>>advantages (lower cost, faster response to bugs, developer
>>enthusiasm). But until Oracle and Informix made their announcements,
>>it was going nowhere. It wasn't until these key pieces of
>>infrastructure existed that other people were willing to take the
>>plunge. But Linux on x86 and Linux on, say, SPARC V9 are two
>>completely different platforms as far as this is concerned. For that
>>matter, Linux/x86 and Linux/PPC are completely different platforms,
>>too.
>>
>>> Furthermore, I think it is quite realistic to expect it to *actually
>>> happen* this way.
>>
>>I don't, because there isn't a critical mass of 64-bit hardware and
>>software, and there isn't the kind of push behind it that there is
>>behind Linux in general.
>
>There are enough Alpha boxes out there that ApplixWare has been deployed
>on Alpha, and there are a good half dozen VARs busy selling Linux/Alpha
>systems, so the situation there is, if not really impressive, certainly
>not really bad.
>
>I certainly agree that Compaq/Digital have done little to capitalize
>on their *huge* lead in "64 bit." They should have been pushing it,
>HARD. It would have been Very Cool if Apple had gone with Alpha
>rather than PPC; it would be Similarly Cool if Amiga were doing the
>same. As it stands, Alpha looks like it will be acting like "deer
>in the headlights" against the onslaught of the Intel marketing
>juggernaut, despite vastly better maturity.
>
>>> The approach appears to conform with Linus' attitudes towards the issue,
>>> which is a rather important issue. If he feels things should be done in a
>>> particular way, that is likely to matter quite a lot. If you think it is
>>> "not realistic," that, frankly, doesn't weigh heavily on peoples' minds.
>>
>>So reality isn't important, only elegance? That's actually rather
>>ironic, given the early Linus-Andy Tannenbaum debate over
>>microkernels.
>>
>>Linus is certainly opposed to changing the ext2 filesystem, which
>>position I agree with; it's currently very stable and there's nothing
>>to be gained by risking that stability or forcing people to flag day
>>their existing filesystems. To the best of my knowledge, though,
>>Linus has no objection to creating new filesystems that can handle
>>64-bit files.
>
>Apparently you still haven't grasped that filesystems are *not the
>problem.*
>
>ext2 already supports 64 bit filesizes quite nicely.
>
>Let me say that again.
>
>ext2 already supports 64 bit filesizes quite nicely.
>
>The *problem* is that the 32 bit VFS and memory manager can't support
>them.
>
>Stephen Tweedie's "ext3" will likely also support 64 bit filesizes,
>on architectures where VFS supports 64 bits.
>
>The same is likely to be true when Reiserfs gets introduced into
>"official" Linux kernel releases.
>
>Filesystems are hardly easy to implement, but their data structures
>are not too difficult to extend. Memory management systems are
>what are hard to rearchitect.
>--
>>Ever heard of .cshrc?
>That's a city in Bosnia. Right?
>(Discussion in comp.os.linux.misc on the intuitiveness of commands.)
>[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: Daniel Forester <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.misc,comp.os.linux.networking,comp.os.linux.hardware,comp.os.linux.development.apps
Subject: Re: Linux Training
Date: 23 Jul 1999 03:26:39 GMT
In comp.os.linux.misc JamesH <[EMAIL PROTECTED]> wrote:
: Can anyone recommend a good Linux Training Organization?
: Has anyone used or heard anything about Linuxcare or Redhat training?
http://www.linuxgeneralstore.com/
(in atlanta... No, I'm not affiliated, nor have I even been to the place
(though I've heard some stuff about 'em, mostly good); just providin' some
sources)
--
Daniel E. Forester
Georgia Institute of Technology
http://www.prism.gatech.edu/~gte061f/
And God said, "Let there be vodka!" And He saw that it was good.
Then God said, "Let there be light!" And then He said, "Whoa -
too much light."
------------------------------
From: jauming <[EMAIL PROTECTED]>
Subject: libc
Date: Fri, 23 Jul 1999 03:26:54 GMT
howdy, gals & guys
we know libc-5.4.46 is for 386/486/586/m68k
is there any newer release that is for "mips"-linux?
thx in advanced!:)
you can also reply to [EMAIL PROTECTED]
--
--
regards
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: "fernando Ortega" <[EMAIL PROTECTED]>
Subject: How to create a module?
Date: Thu, 22 Jul 1999 19:43:49 +0200
I�ve been reading some doc�s about programming modules, but I got confused.
What can I consider to be a module exactly?
Can a file of the kernel be consider as a module?
For example if I want to modify just a file of the whole kernel, can I
convert that file in a module?
If so please, would you be so kind to tell me , HOW?
Thank you very much in advance, Fernando Ortega
------------------------------
From: [EMAIL PROTECTED] (Don Scheidt)
Subject: Socket connection to Serial I/O - How?
Date: Fri, 23 Jul 1999 04:14:02 GMT
Seeing the good stuff about sockets posted here, I thought I'd
pose a similar question. Rather than go out and hook up to a
TCP/IP point, I'm trying to do a socket-type thing that would
connect up to, say, /dev/ttyS0 (aka "Com1"). The other end
would be bound (am I using that term correctly) to a process ID
in /var/tmp (I think... that's what I'm working with right now,
anyway). The idea is to get a server that accepts scripts to
talk via this socket to /dev/ttyS0, which would have some clever
little device on it that would accept simple commands and send
back responses over the same socket-connection. I'm still a
touch confused about doing this, even after reading some of
the very good documentation out there on sockets, mostly because
I'm seeing more TCP/IP connection stuff, rather than the
relatively simple (?) type of "app" I'm trying to cob together.
Soooo... any hints from the gurus and enlightened programmers
would be helpful - as always, a code fragment or two in the
right direction would work wonders, too. I'm having great
good fun doing this, and sucess would be fine thing to
acheive... thanks for any help.
--
Don Scheidt, [EMAIL PROTECTED]
------------------------------
From: "Margaret Dailey" <[EMAIL PROTECTED]>
Crossposted-To:
alt.comp.lang.visualbasic.ver3,comp.arch.embedded,comp.arch.fpga,comp.dsp,comp.lang.ada,comp.os.vxworks,comp.realtime
Subject: Re: "Contract Outsourcing?!"
Date: Fri, 23 Jul 1999 20:45:37 -0700
Aww, Blake, no fair. How come you don't like us? Dare I ask?
------------------------------
From: Helmut Artmeier <[EMAIL PROTECTED]>
Subject: Linux hardware debugging
Date: Fri, 23 Jul 1999 07:46:31 +0200
Hi!!!
I have downloaded Dag Brattli's irdatools. I want to use irdadump but
the debug information out of irdadump are very confusing:
09:02:09.620264 xid:cmd 0xffffffff < 0xd152961e S=6 0
09:02:09.690178 xid:cmd 0xffffffff < 0xd152961e S=6 1
09:02:09.750173 xid:cmd 0xffffffff < 0xd152961e S=6 2
09:02:09.820145 xid:cmd 0xffffffff < 0xd152961e S=6 3
09:02:09.880370 xid:cmd 0xffffffff < 0xd152961e S=6 4
09:02:09.880676 xid:rsp 0x00005690 > 0xd152961e S=6 4 LnxMachine
hint=0d00 [ PnP Computer Printer ]
09:02:09.970153 xid:cmd 0xffffffff < 0xd152961e S=6 5
09:02:10.060239 xid:cmd 0xffffffff < 0xd152961e S=6 * Sim1-Client
hint=8404 [ Computer IrCOMM ]
09:02:10.100803 snrm:cmd ca=0xfe pf=1 0x00005690 < 0xd152961e
new-ca=0xe6
09:02:10.101164 ua:rsp ca=0xe6 pf=1 0x00005690 > 0xd152961e
09:02:10.101349 ua:rsp ca=0xe6 pf=1 0x00005690 > 0xd152961e
09:02:10.410419 rr:cmd < ca=0xe6 pf=1 nr=0
09:02:10.410742 rr:rsp > ca=0xe6 pf=1 nr=0
09:02:10.430366 i:cmd < ca=0xe6 pf=1 nr=0 ns=0 LM slsap=0x03 dlsap=0x00
CONN_CMD
09:02:10.430721 rr:rsp > ca=0xe6 pf=1 nr=1
09:02:10.450332 rr:cmd < ca=0xe6 pf=1 nr=0
09:02:10.450621 rr:rsp > ca=0xe6 pf=1 nr=1
09:02:10.470325 rr:cmd < ca=0xe6 pf=1 nr=0
09:02:10.470615 rr:rsp > ca=0xe6 pf=1 nr=1
09:02:10.490329 rr:cmd < ca=0xe6 pf=1 nr=0
...
09:02:30.870353 i:cmd < ca=0x4e pf=1 nr=0 ns=0 LM slsap=0x03 dlsap=0x00
CONN_CMD
...
09:02:54.890366 disc:cmd < ca=0xe0 pf=1
09:02:54.890660 ua:rsp ca=0xe0 pf=1 0x00005690 > 0xd152961e
09:02:55.800481 snrm:cmd ca=0xfe pf=1 0x00005690 < 0xd152961e
new-ca=0x8e
09:02:55.800882 ua:rsp ca=0x8e pf=1 0x00005690 > 0xd152961e
09:02:55.801073 ua:rsp ca=0x8e pf=1 0x00005690 > 0xd152961e
09:02:56.110344 rr:cmd < ca=0x8e pf=1 nr=0
09:02:56.110651 rr:rsp > ca=0x8e pf=1 nr=0
09:02:56.130344 i:cmd < ca=0x8e pf=1 nr=0 ns=0 LM slsap=0x03 dlsap=0x00
CONN_CMD
Maybe you know something about the possible debugging information out of
irdadump...
Helmut Artmeier
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Philipp Thomas)
Subject: Re: Asynchronous IO networking Stack?
Date: Fri, 23 Jul 1999 03:28:51 GMT
On Sun, 18 Jul 1999 20:28:07 +0800, XuYifeng <[EMAIL PROTECTED]> wrote:
>WSAWaitForMultipleEvents only allow wait on 64 objects, if you have 65
>objects, how do you wait on them?
Just on a side note, if you really need more than 64 objects you have
some serious flaws in your design.
Philipp
--
Close the windows! The penguin is freezing.
------------------------------
** 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
******************************
