Linux-Development-Sys Digest #78, Volume #7 Fri, 20 Aug 99 21:14:32 EDT
Contents:
Re: why not C++? (Randall Parker)
Re: why not C++? (Randall Parker)
Re: X Windows developement (Dave Newton)
Re: Troll (was: why not C++?) (Johan Kullstam)
Direct programming of /dev/eth device? ("Richard Krehbiel")
Re: Custon Device Drivers (Grant Edwards)
Help on customizing libc/libgcc/libg++ startup code (Marco Zandonadi)
Re: libstdc++.2.8 problem (David T. Blake)
Re: libstdc++.2.8 problem (Allin Cottrell)
Re: Having a problem loggin in, need help (Yannick Patois)
Re: Why not C++ (Kaz Kylheku)
Re: SCSI Tekram DC-390F on linux 2.0.36 (or 2.2.11) (David J. MacKenzie)
3com ISA cards and linux ("Kalkas")
Re: Having a problem loggin in, need help (Eric Hegstrom)
Re: Why so inefficient source RPM's ?? (Piercarlo Grandi)
libstdc++.2.8 problem (Tom Daley)
Re: Lisp OSes (was: Re: Troll (was: why not C++?)) ("Noah Roberts (jik-)")
Re: 3com ISA cards and linux ("Caitanya")
Re: 3com ISA cards and linux (Toni Grass)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Randall Parker)
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: why not C++?
Date: Fri, 20 Aug 1999 13:18:40 -0700
In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED]
says...
> MHO: optimizations for specific hardware should be the responsibility of
> the compiler
But if the compiler doesn't do the optimization the result is slower
code.
If, for instance, you don't need ii to go from 0 to 9 and it is just as
easy to count down then it makes sense to code your for loops that way.
Almost all processor instruction sets have a decrement and test for 0 and
by doing the decrement to 0 you may avoid the reloading of the boundary
test value at the bottom of pass thru a big loop. I fail to see how such
an optimization is going to slow it down for some processors so why not
make it a standard practice?
I've been faced with situations where my choice was either write it in
assembly to make it fast enough or write very careful C to make it fast
enough. Well, I chose the latter.
One of the advantages of your coding carefully in C is that you rarely
will thereby write code that an optimizer will make go slower and usually
even when using an optimizing compiler the code will be faster than code
written with less thought of the fact that underlying hardware will
execute it.
>and, given that you're interested in writing general,
> portable code and not device drivers, etc. the programmer should not
> bother to concern him/herself with them.
Most C optimizations are not specific to a particular processor
architecture. Most of those that are are not, as a result, going to be
slower when later compiled for a different processor architecture.
However, the fact is that if you are writing in C and you do not
consciously code for a particular processor architecture your style may
accidentally lend itself to one processor architecture over another.
For instance, some processor architectures can shift left faster than
shifting right (or maybe I have that backward). So if you do your
shifting always one way you are writing your code to be faster or slower
for a particular processor architecture whether you are aware of this or
not.
> I never look at what my code is making the machine actually do when
> writing C, since I write code that runs on many platforms and that kind
> of optimization in C is useless (or worse) in the large.
You ought to start looking at the generated assembly on a couple of
different architectures and see whether your carefree attitude isn't
consistently costing you performance loss unnecessarily.
> If I care that much about what the machine is actually doing, well,
> that's what the asm statement is for, right?
If you care enough about what the machine is doing you can code C better
so that the resort ot asm can happen less often.
------------------------------
From: [EMAIL PROTECTED] (Randall Parker)
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: why not C++?
Date: Fri, 20 Aug 1999 13:06:39 -0700
It is not a concocted problem.
I've run into this sort of situation when dealing with writing interfaces
to a set of embedded devices which each had slightly different ways of
formatting their packets for serial communication.
I solved the problem by declaring pointers to functions where each
function had a different way of formatting output or of interpreting
input. I could make up a description for how to communicate to a
particular embedded device by selecting a set of pointers that together
worked with that device.
I've run into this kind of situation in other problem domains as well.
Lots of people solve it by putting in lots of flags and conditions within
code that gets executed so that in every pass thru the code the
particular way of doing each step has to be selected for every time thru.
Well, I don't like testing flags more often than necessary. It strikes me
as wasteful. Also, each routine that does get executed ends up being
bigger than it needs to be and more code has to be loaded into cache.
I've worked with embedded devices where the CPU had a large L1 cache and
all external RAM had many wait states on it. In those cases the ability
to make things small enough to fit in cache pays big dividends in
performance.
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] says...
>
> Agreed that the language does not provide the capability you describe.
> However, one can easily achieve what you want.
>
> Let's call the base class MyClass1 and the subsequent classes MyClass2,
> MyClass3, and MyClass4 respectively. The implementation of MyClass4::bb
> would look like the following.
>
> ReturnType MyClass4::bb(Args args)
> {
> return MyClass1::bb(args);
> }
>
> This seems to be a more than adequate answer to your need, I think.
>
> I am curious to know whether this is a concocted problem or have you
> really faced such a problem?
>
>
>
------------------------------
From: Dave Newton <[EMAIL PROTECTED]>
Subject: Re: X Windows developement
Date: Fri, 20 Aug 1999 19:59:16 GMT
Tranceport <[EMAIL PROTECTED]> wrote:
> I have to develop Xaw (X athena widgets, i think)
I gotta ask... why?!?!
You'll need to look at the source and the old Xlib
programmer's manual, IIRC. (Or Xt, it's been awhile.)
Dave
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: Johan Kullstam <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: Troll (was: why not C++?)
Date: 19 Aug 1999 09:35:29 -0400
Stephan Houben <[EMAIL PROTECTED]> writes:
> [EMAIL PROTECTED] (Rainer Joswig) writes:
>
>
> > A Lisp users says: I don't care what language the kernel
> > is written in, as long as it is Lisp.
>
> IMHO, a supposedly "general purpose OS" should do more than
> just please the Lisp users. (Any estimates of the precentage
> of the programming community which consider themselves
> Lisp users?)
popularity means nothing. look how many windows users there are.
compare it to the number of linux users. is windows all that great?
does linux suck so much?
lisp isn't good for everything, but it is a good language for
designing a GUI. imho C++ is awful for this task. but what is
everyone using? the bandwagon effect doesn't select for quality.
> > The difference between a Lisp machine and a Linux
> > system is huge. The Lisp Machine has an open, incrementally
> > changeable and dynamic OS. The OS is written in a
> > pure OOP style. That means that things like
> > IP-Packets or Processes are objects. And operations
> > on them are implemented as methods.
>
> But all this could be provided by a Lisp library which interfaces
> with the underlying OS. Have the Linux kernel start up a Lisp
> interpreter directly after booting and presto: your Lisp OS is
> ready to run. You can then write a GUI or whatever in Lisp and
> have *that* executed on startup.
try to use linux without a libc. everything in linux goes to support
a C environment. static compile languages such as pascal or fortran
do well on unix. dynamic languages such as lisp do less well. this
is because a lisp environment has aspirations of being an operating
system itself and the kernel/app domains overlap.
the kernel of a lisp oriented box could be written in another language
than lisp such as C. however, it would offer a different set of
interfaces and services than unix. what lisp needs from its kernel is
different from what a C program needs.
> Of course, the advantage that this gives you from just running
> a Lisp interpreter from the shell is debatable. At least, you can
> pretend you're not using Linux...
as far as i understand it, it is a world of difference. a lisp isn't
some kind of scripting application. it's nearly a full blow operating
system with compiler, memory management, subprocesses &c. look at
emacs. emacs isn't great lisp, but it sure wants to take over the
world.
--
johan kullstam
------------------------------
From: "Richard Krehbiel" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: Direct programming of /dev/eth device?
Date: Thu, 19 Aug 1999 09:11:02 -0400
I'm about to embark on a project that intends to translate a very
proprietary in-house ethernet protocol into TCP/IP. My current research
tells me that a DOS packet driver interface is capable of doing what I need,
so I've been proceeding in that direction.
But you know, I'd much prefer to do this on Linux.
Can I get my hands dirty with ethernet devices? I need to program a
specific ethernet address, I need to send and receive on a specific ethernet
multicast address, and of obviously I need to send and receive packets.
Thanks.
--
Richard Krehbiel, Arlington, VA, USA
[EMAIL PROTECTED] (work) or [EMAIL PROTECTED] (personal)
------------------------------
From: grant@nowhere. (Grant Edwards)
Subject: Re: Custon Device Drivers
Date: Thu, 19 Aug 1999 17:37:08 GMT
In article <[EMAIL PROTECTED]>, Kaz Kylheku wrote:
>>We have developed an ISA card that is only avalable from us. If I was
>>to write a device driver for this card would I have to publish the
>>source for the device driver? or can we maintain that the binary is
>What the binary form does is make it difficult for people to reverse engineer
>the code, and next to impossible for them to patch bugs or make adaptations.
In other words, you're turning away free engineering help from
a group of very talented people. Of course if your board is
only purchased by people who don't know anything about
software, then maybe you're not really loosing out.
>As a hardware maker, why would you care?
It could be that most of the "value" they're selling is in the
driver, not in the hardware. If they publish sources to the
driver, somebody else can come up cheap, compatible hardware,
and then they've got no product.
>Don't you want people to understand
>the hardware interfaces so that the hardware will achieve broad support? same
>interface and then just use your driver?
--
Grant Edwards grante Yow! I will SHAVE and
at buy JELL-O and bring my
visi.com MARRIAGE MANUAL!!
------------------------------
From: Marco Zandonadi <[EMAIL PROTECTED]>
Crossposted-To: gnu.gcc.help,comp.os.linux.development.apps
Subject: Help on customizing libc/libgcc/libg++ startup code
Date: Fri, 20 Aug 1999 13:49:24 -0700
Hello,
I need some pointers on how to customize the libc, libgcc and libg++
startup code both with gcc/g++ and with egcs.
Any help would be greatly appreciated.
Marco Zandonadi
------------------------------
From: [EMAIL PROTECTED] (David T. Blake)
Subject: Re: libstdc++.2.8 problem
Date: 20 Aug 1999 21:02:05 GMT
Reply-To: [EMAIL PROTECTED]
Tom Daley <[EMAIL PROTECTED]> wrote:
> I am having problems building a usable libstdc++.so.2.8.X
>
> The other day I got Netscape 4.61 glibc version. It would not run
> because:
...
> I started with slackware 4.0, and have since updated to glibc-2.0.7pre6
> (compiled with egcs-1.1.1) and recently updated the compiler to gcc-2.95.
>
> Can anyone help me?
Yes.
Update to glibc2.1, but ONLY AFTER READING AND RE-READING THE
FAQ FOR UPGRADING glibc. Hint: You are likely to break just
about everything on your system if you don't read the upgrade
documents before proceeding.
--
Dave Blake
[EMAIL PROTECTED]
------------------------------
From: Allin Cottrell <[EMAIL PROTECTED]>
Subject: Re: libstdc++.2.8 problem
Date: Fri, 20 Aug 1999 17:22:55 -0400
Tom Daley wrote:
>
> I am having problems building a usable libstdc++.so.2.8.X
>
> The other day I got Netscape 4.61 glibc version. It would not run
> because [of the usual stuff]
Try grabbing an appropriate lib from
ftp://ricardo.ecn.wfu.edu/pub/netscape-libs
--
Allin Cottrell
Department of Economics
Wake Forest University, NC
------------------------------
From: Yannick Patois <[EMAIL PROTECTED]>
Subject: Re: Having a problem loggin in, need help
Date: Fri, 20 Aug 1999 18:56:15 +0000
Advanced one corporation wrote:
>
> I'm having a problem loggin in. After I have logged in to my localhost by
> typing in root and then my password, this message appears:
> [root@localhost /root#
type :
# shutdown -h now
Go to the nearest bookstore buy a unix book and read it, then switch
your computer on again.
Yannick
--
_/ Yannick Patois \___________________________________________________
| irc(undernet): Garp on #france25+ | CALVIX ! Le LUG de CAEN |
| web :http://www.sura.org/~patois/ | web : http://calvix.sicfa.org/ |
| email : [EMAIL PROTECTED] | |
|___________________________________|__________________________________|
------------------------------
From: [EMAIL PROTECTED] (Kaz Kylheku)
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.networking
Subject: Re: Why not C++
Date: Fri, 20 Aug 1999 18:56:03 GMT
On 20 Aug 1999 09:25:08 -0400, Johan Kullstam <[EMAIL PROTECTED]> wrote:
>and it would be nice to be able put in a goto of some sort to do
>something useful.
>
> writeh(array, size, fd, goto disk_full);
>
>and failed writes could jump to some recovery code.
The above might not be C syntax, but a macro can give it to you.
I've maintained code in which REQUIRE() macros were used heavily to do this
kind of thing. These macros would jump to a given statement label if a certain
condition were found to be false, and log a message. These were only used
because the code had to work in C++ environments that do not support exception
handling.
C gives you setjmp and longjmp that can be used to implement exception handling
that is quite functional. I have written a tiny package which can give C
programs structured exception handling. You can have nested try blocks, and
exceptions that are divided into groups so you can catch by group as well as
specific ID. The calling of cleanup handlers is supported in the unwinding
process. Unhandled exceptions go to a default handler which can be overriden
by a pointer to a user-supplied function.
>it might be nice to have functions check for stuff automatically and
>by default unless told otherwise. then it wouldn't be such a royal
>pain. computers are great at filling the details. they don't get
>tired of repitition. let the computer do the religious checking!
This is achieved to some extent exception handling. In C++, operator new now
is required to throw an exception if allocation fails, unless the nothrow form
is used. If you do not handle the exception, the program is aborted. You can
implement a function called terminate() which catches unhandled exceptions
which can do things other than terminate; for example, re-execute a new
program image. Of course, you do not have a simple compile time switch which
will turn off all your try blocks and let all exceptions be unhandled.
If you want to write a program that does little error checking, just have the
low level functions throw exceptions that are not caught. Without too many
problems, you can implement your own I/O layer which wraps up some
non-exception based I/O functions and translates errors into exception throws.
Such programs would traditionally be peppered with nasty exit calls. In
reality, only certain classes of programs can be written this way, such as
small trivial programs or batch processors such as compilers, which can abort
at the first resource problem or I/O error. In many types of applications,
termination is an unacceptable way of dealing with transient environmental
errors. And even if a shutdown must take place due to a catastrophic failure,
it may be required that resources must be released and everything must be left
in a sane state. E.g. a database would probably want to close all of its tables
properly, or a program with resources that are not automatically deallocated by
the OS would probably want to free them.
------------------------------
From: [EMAIL PROTECTED] (David J. MacKenzie)
Crossposted-To: comp.os.linux.hardware
Subject: Re: SCSI Tekram DC-390F on linux 2.0.36 (or 2.2.11)
Date: 16 Aug 1999 22:06:21 -0400
Dorin-Ioan MARINCA <[EMAIL PROTECTED]> writes:
> What I must check on "make -xconfig" windows to create an linux kernel
> (2.0.36 or 2.2.10/11) which supports an SCSI controller Tekram DC-390F?
Probably use the NCR53C8XX driver. I'm using it right now on 2.2.5
with the top-end Tekram 390 series controller, I forget the model #.
------------------------------
From: "Kalkas" <[EMAIL PROTECTED]>
Crossposted-To:
at.linux,aus.computers.linux,be.comp.os.linux,comp.os.linux,comp.os.linux.advocacy,comp.os.linux.hardware,comp.os.linux.misc,comp.os.linux.networking
Subject: 3com ISA cards and linux
Date: Sat, 21 Aug 1999 02:19:56 +0200
I have been seriously thinking to use Linux and stop using Windows 98. I am
fascinated by Linux's stability and security.
Therefore, I have seriously planing to install Linux and USE IT.
However, it seems that it is not possible for me to use Linux, since I use
cable modem with a 3com ISA card. More precisely, I use 3com EtherLink III
ISA (3C509/3C509b) network interface card, and there are no drivers which
will support my card in Linux.
Did someone else have similar problems?
Regards,
Kalkas
------------------------------
From: Eric Hegstrom <[EMAIL PROTECTED]>
Subject: Re: Having a problem loggin in, need help
Date: Fri, 20 Aug 1999 12:12:30 -0700
try tying "startx" to (hopefully) load up some sort of desktop so you
can play.
-Eric
Advanced one corporation wrote:
>
> I'm having a problem loggin in. After I have logged in to my localhost by
> typing in root and then my password, this message appears:
> [root@localhost /root#
>
> What does this message mean and what do I type.
> Help is appreciated.
> Thanks
> [EMAIL PROTECTED]
--
Eric Hegstrom .~.
Senior Software Engineer /V\
Sonoran Scanners, Inc. // \\ L I N U X
[EMAIL PROTECTED] /( )\ >don't fear the penguin<
520-617-0072 x402 ^^-^^
------------------------------
From: [EMAIL PROTECTED] (Piercarlo Grandi)
Crossposted-To: linux.redhat.misc,linux.redhat.rpm
Subject: Re: Why so inefficient source RPM's ??
Date: 20 Aug 1999 20:25:08 +0100
>>> On 16 Aug 1999 23:32:20 +0200, Peter Mutsaers <[EMAIL PROTECTED]> said:
Peter> Coming from FreeBSD, I'm used to the ports system. It is a bit
Peter> like a source RPM: you have an original tarball, patches, build
Peter> instructions and a file list. This is handled automatically, then
Peter> packaged and installed.
Peter> But, when someone does a minor adjustment to the port, you don't
Peter> have to download again the whole (sometimes very large) original
Peter> source, only the update patch or build instructions, usually very
Peter> small and thus it is very easy and efficient to stay current
Peter> (even more so because of the excellent cvs/cvsup system). [ ... ]
Yeah, this annoys me too. However it is easy to explain: RPM design was,
and is, still dominated by the specific needs/incentives of RedHat
Inc. (down to the inane "/usr/src/redhat" structure): since RedHat SRPMS
come neatly on a CD when you buy an official RedHat boxed set, there is
absolutely no need to make downloading of SRPMSes faster.
I'd love too to have SRPMSes by default not incorporate the original
source archive, and download it separately; this can even be done, and
has been done, for packages whose source cannot be redistributed by
RedHat. But the default is still what suits the RedHat way of doing
things.
------------------------------
From: [EMAIL PROTECTED] (Tom Daley)
Subject: libstdc++.2.8 problem
Date: 20 Aug 1999 19:19:35 GMT
I am having problems building a usable libstdc++.so.2.8.X
The other day I got Netscape 4.61 glibc version. It would not run
because:
./netscape: error in loading shared libraries
libm.so.5: cannot open shared object file: No such file or directory
But "ldconfig -v" shows that libm.so.5 is in it's path.
root@penguin /usr/lib >ldconfig -v
ldconfig: version 1.9.9
/usr/i486-linuxlibc5/lib:
libc.so.5 => libc.so.5.4.46
libm.so.5 => libm.so.5.0.9
/usr/local/lib:
<snip>
So I added "/usr/i486-linuxlibc5/lib" to my LD_LIBRARY_PATH and then
netscape gets a seg fault.
(gdb) where
#0 0x40161ae5 in _IO_file_attach () from /usr/lib/libstdc++.so.2.8
#1 0x401e0805 in _IO_fdopen (fd=6, mode=0x8a40e6d "w") at iofdopen.c:116
#2 0x83dead9 in DNS_SpawnProcess ()
#3 0x82bb7bc in XFE_InitDNS_Early ()
#4 0x82a7ca4 in main ()
Also, I ran ldd on /usr/lib/libstdc++.so.2.8 and saw that it was the file
that needed libm.so.5.
I decided to try and rebuild libstdc++.so.2.8. I got gcc-2.8.1 and the
source for libstdc++-2.8.1.1. I built and installed libstdc++.so.2.8.1.1
and now netscape reports:
./netscape: error in loading shared libraries
/usr/lib/libstdc++.so.2.8: undefined symbol: _vt.7filebuf
I am not sure what to try next.
I started with slackware 4.0, and have since updated to glibc-2.0.7pre6
(compiled with egcs-1.1.1) and recently updated the compiler to gcc-2.95.
Can anyone help me?
Thanks
--
===================================================================
| o Tom Daley |
| ___ </v Woodland Park, CO |
| ___ -\ [EMAIL PROTECTED] |
| ___ / (719) 534-0449 x27 |
| (*) Linux! |
===================================================================
------------------------------
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: Lisp OSes (was: Re: Troll (was: why not C++?))
From: "Noah Roberts (jik-)" <[EMAIL PROTECTED]>
Date: 20 Aug 1999 16:56:49 -0700
[EMAIL PROTECTED] (Christopher B. Browne) writes:
> On 20 Aug 1999 08:59:59 +0200, Stephan Houben
> <[EMAIL PROTECTED]> posted:
> >OK, you have me convinced. I want it.
> >
> >Is anyone working on a new Lisp OS nowadays?
>
> See: <http://www.hex.net/~cbbrowne/lisposes.html> for links to several
> projects.
That site is quite broken...one link goes to a credit card sales pitch!
>
> It's not clear which, if any, will get sufficiently developed as to be
> of real interest...
>
> --
> "Note that if I can get you to `su and say' something just by asking,
> you have a very serious security problem on your system and you should
> look into it." -- Paul Vixie, vixie-cron 3.0.1 installation notes
> [EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: "Caitanya" <[EMAIL PROTECTED]>
Crossposted-To:
at.linux,aus.computers.linux,be.comp.os.linux,comp.os.linux,comp.os.linux.advocacy,comp.os.linux.hardware,comp.os.linux.misc,comp.os.linux.networking
Subject: Re: 3com ISA cards and linux
Date: Sat, 21 Aug 1999 10:59:14 +1000
I'm on cable as well the guys are great there and you can joing the bigpond
linux newsgroup, read the FAQ and you'll get up and running in not time at
all
============================================================================
=======================================
Kalkas wrote in message <[EMAIL PROTECTED]>...
I have been seriously thinking to use Linux and stop using Windows 98. I am
fascinated by Linux's stability and security.
Therefore, I have seriously planing to install Linux and USE IT.
However, it seems that it is not possible for me to use Linux, since I use
cable modem with a 3com ISA card. More precisely, I use 3com EtherLink III
ISA (3C509/3C509b) network interface card, and there are no drivers which
will support my card in Linux.
Did someone else have similar problems?
Regards,
Kalkas
------------------------------
From: Toni Grass <[EMAIL PROTECTED]>
Crossposted-To:
at.linux,aus.computers.linux,be.comp.os.linux,comp.os.linux,comp.os.linux.advocacy,comp.os.linux.hardware,comp.os.linux.misc,comp.os.linux.networking
Subject: Re: 3com ISA cards and linux
Date: Sat, 21 Aug 1999 00:46:51 GMT
Kalkas wrote:
>
> I have been seriously thinking to use Linux and stop using Windows 98. I am
> fascinated by Linux's stability and security.
>
> Therefore, I have seriously planing to install Linux and USE IT.
>
> However, it seems that it is not possible for me to use Linux, since I use
> cable modem with a 3com ISA card. More precisely, I use 3com EtherLink III
> ISA (3C509/3C509b) network interface card, and there are no drivers which
> will support my card in Linux.
>
> Did someone else have similar problems?
>
This card is definitely supported by Linux!! I used the same NIC
together with a cable modem. The only problem is plug&punish - you will
have to switch it off.
regards
Toni
------------------------------
** 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
******************************
