Linux-Development-Sys Digest #556, Volume #6 Wed, 31 Mar 99 15:14:46 EST
Contents:
Re: 2.2.[2,3] cannot mount MO devices (Mike Dowling)
'ping -f' causes Tulip to hang (kernel 2.2.5) (BL)
Re: clone() or PThreads ??? ("Udo Giacomozzi")
development features ("Jeff D. Hamann")
compiling problem ("Rick Gocher")
Re: Help me with my CGI-prog ("Sascha Bohnenkamp")
Re: problems removing linux from my Hard Disk (Frans de Wet)
Re: 4 Gb memory? (Robert Krawitz)
Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC. (Joseph H Allen)
accessing a filesystem out of a module (Florian Erhard)
Re: JDBC driver and JDBC-ODBC bridge ("Darrel Davis")
Keyboard and Mouse (Aries)
Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC. (Donal K.
Fellows)
Re: [ANN] CodeWarrior for Red Hat Linux ("Paul F. Dietz")
Re: CodeWarror for Linux (was: Re: Programming tools for ...) (Klaus Schilling)
Re: clone() or PThreads ??? ("G. Sumner Hayes")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Mike Dowling)
Subject: Re: 2.2.[2,3] cannot mount MO devices
Date: 31 Mar 1999 15:48:56 GMT
On Wed, 31 Mar 1999 10:21:09 +0000, Igor Zlatkovic <[EMAIL PROTECTED]> wrote:
>Just did it.
>
>I can confirm that I am unable to mount the MO drive under 2.2.5. The MO diskette was
>sucessfully formatted under 2.2.1, some data was copied to it, rebooted to 2.2.5 and
>mount does not work. It says that it cannot find ext2 filesystem on the device.
>
Thanks, I'm relieved that somebody else is experiencing this too.
In my stupidity, I clobbered my old 2.2.1 kernel, which puts me rather in a
spot!
Cheers,
Mike Dowling
--
My email address [EMAIL PROTECTED] above is a valid email address.
It is, in fact, a sendmail alias; the digit 'N' is incremented regularly.
Spammed aliases will be deleted. Currently, mike[5,7,8] have been deleted.
If email to mikeN bounces, try mikeN+1.
------------------------------
Crossposted-To: comp.os.linux.networking
From: BL <[EMAIL PROTECTED]>
Subject: 'ping -f' causes Tulip to hang (kernel 2.2.5)
Date: Wed, 31 Mar 1999 15:46:29 GMT
I have 2 linux boxes connected via Tulip cards. one box is 2.0.36 (a linux
'router') and the other is kernel 2.2.x (I've tried all 'x' versions up to
latest, 5).
the connection is via a rolled (xover) cable. direct pt-pt.
whenI start the 'ping -f -s 1000' command on the 2.2 system, it seems to work
for about 30 seconds or so, then the backspaces (meaning the echo replies)
don't print and it seems the card then chokes and hangs. from then on, the
card or system won't respond. turning off network and back on again (etc/rc.d
stuff) doesn't help. only a reboot helps.
note that when the 2.2 system reboots, its again reachable from the 2.0.36
system. the 2.0.36 system does not appear to need a reboot to fix the hang.
anyone know of any network saturation issues with the 2.2 kernel series?
when I get a chance, I boot both systems to 2.0.36 and test some more. but
the 2.2 system is running dual cpu's (would that have anything to do with it?)
and I'd like to stay at 2.2 if possible.
the 'router' is a k6-2 system and doesn't really need the 2.2 kernel. and its
my entrypoint to the internet so it has to be stable ;-)
thanks,
--
bryan
EMAIL--> bryan [at] grateful (dot) net
------------------------------
From: "Udo Giacomozzi" <[EMAIL PROTECTED]>
Subject: Re: clone() or PThreads ???
Date: Wed, 31 Mar 1999 14:52:16 +0200
First, thank you for your interesting reply!
Let me tell you more about my application.
As I already said, I need to separate the DSP routines from the GUI. I
probably will need other child processes for background works, but that's
another question.
It seems I need to use fork() because the RTL is not reentrant. Ok.
The reason why I want to share the memory is because my application is made
completely object-oriented, including sound routines. The class functions
and data should be accessible from both processes. I don't think this will
be a problem because class functions are reentrant. I don't know if I really
need shared memory. All I need is access to the classes (owned by the
parent). So if I can pass the pointer to the class to the child and use it,
there is no problem (no shared memory is needed). But I actually don't know
if this is possible under Linux (as memory management is different than
under Windows).
So, what do you recommend me?
Regards
Udo Giacomozzi
--
* Email: [EMAIL PROTECTED]
* UIN: 17745247 (@pager.mirabilis.com)
------------------------------
From: "Jeff D. Hamann" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: development features
Date: Wed, 31 Mar 1999 09:07:20 -0800
I have been developing Windows software for some time and
have been considering moving to Linux. I develop apps
that utilize ODBC, sockets, and multithreading. I know that
Linux can handle the latter two, but I would like to find out
what database capabilites exist for interfacing with
Windows ODBC. Also I have grown accustomed to the
common controls and like what I see in GNOME and KDE.
Is there a site that has the documentation for the
class libs for the two?
Jeff.
--
Jeff D. Hamann
277 SE Lilly
Apartment H
Corvallis, Oregon 97333-1870 USA
541-754-3309
[EMAIL PROTECTED]
http://www.osu.orst.edu/~hamannj
begin 666 Jeff D. Hamann.vcf
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:.3 S,S%4,3<P-S$Y6@T*14Y$.E9#05)$#0H`
`
end
------------------------------
From: "Rick Gocher" <[EMAIL PROTECTED]>
Subject: compiling problem
Date: Wed, 31 Mar 1999 17:29:57 GMT
Hi all,
I have been trying to compile ssh for my RH5.2 box. It keeps ending with
several errors about undefined references to tgetstr on a file called
shreadline.c? Can anyone tell me what tgetstr is how I could fix this?
Thanks for your help,
Rick
------------------------------
From: "Sascha Bohnenkamp" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.programming
Subject: Re: Help me with my CGI-prog
Date: Wed, 31 Mar 1999 15:36:00 +0200
> I've made a little CGI-program that is ran from .shtml. It outputs
>only one line, but the problem is with that it's really made for:
> It keeps track of visitor and show times and such and makes a
>log. However according to my logfile it records only about half of the
>visitors our counter does. Could the problem be with the file system
>i'm using (fopen "a" and fprintf) if there's many hundreds of
> users online at the same time? Please help someone.
>
> Oh, I'm programming with C.
mmh you seem to do concurrent writes with several programms accounting the
users ... you have to do some locking to prevent the programms to overwrite
each
others data etc.
------------------------------
From: [EMAIL PROTECTED] (Frans de Wet)
Subject: Re: problems removing linux from my Hard Disk
Date: Wed, 31 Mar 1999 17:50:23 GMT
Just type
fdisk /MBR
after you boot to the command prompt with your favourite Win95 or DOS
boot disk ... that should do it ..
Frans
On Wed, 31 Mar 1999 13:46:22 +0100, "David Bryan"
<[EMAIL PROTECTED]> wrote:
>I install linux red hat 2.2 onto my PC.
>
>But now I want to remove all traces of linux.
>
>I remove all partions fine, but I can not get rid of the boot manager it
>install
>
>Does any one know how I can remove it??
>
>David Bryan
>
>
------------------------------
From: Robert Krawitz <[EMAIL PROTECTED]>
Subject: Re: 4 Gb memory?
Date: 31 Mar 1999 09:43:09 -0500
"G. Sumner Hayes" <[EMAIL PROTECTED]> writes:
> Think about the memory model for a minute -- processes need a part of
> the address space for themselves. The current default is 1GB kernel
> memory mappings, 3GB maximum process size (having maxprocsize >
> maxramsize makes sense when you consider virtual memory. Ingo's
> patches make it an even 2/2 split. A 3/1 split might be possible but
> probably not very useful; it's not as straightforward as going 2/2 as
> you need to ensure that there aren't places in the kernel which assume
> that you can map all the RAM into one process.
Actually, the 4 GB virtual address space is split between the kernel
(1G), physical memory (1G), and process address space (2G). The
kernel maps all of physical memory into virtual memory so that it can
always directly access all of physical memory. Ingo's patch allows 2G
of physical memory and 1G of process address space.
--
Robert Krawitz <[EMAIL PROTECTED]> http://www.tiac.net/users/rlk/
Tall Clubs International -- http://www.tall.org/ or 1-888-IM-TALL-2
Member of the League for Programming Freedom -- mail [EMAIL PROTECTED]
"Linux doesn't dictate how I work, I dictate how Linux works."
--Eric Crampton
------------------------------
Crossposted-To: comp.os.linux.development.apps,comp.unix.programmer
From: [EMAIL PROTECTED] (Joseph H Allen)
Subject: Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC.
Date: Wed, 31 Mar 1999 18:55:19 GMT
In article <7dsrer$[EMAIL PROTECTED]>,
Alexander Viro <[EMAIL PROTECTED]> wrote:
>In article <[EMAIL PROTECTED]>,
>Joseph H Allen <[EMAIL PROTECTED]> wrote:
>>Why are you guys so down on JOE as a programmer's editor? Just what do you
>>think JOE is missing? I especially don't understand how you think emacs is
>>more keyboard efficient.
>Ability to say c%foo^[ and replace the bracketed subexpression with foo is
>nice. Especially so since d% will delete it, y% - copy to the buffer and %
>move you to the other bracket. The last one is ^Q[ (IIRC), first 3 simply
>do not exist. Yeah, delete a word is ^T and move forward one word is ^F.
In joe it's actually ^G (joe is not wordstar, but 'jstar' is). But yes, joe
does not have the equivalent of vi's 'd' and 'c' commands. 'c' would be
redundant if vi wasn't modal. 'd' is nice, but it has a cost: vi lacks
single key delete commands. Joe has ^D, ^W, ^O, ^Y, ^J, and ^H (I've always
wanted the reverse direction equivalent of ^J on a single key to complete
the set, but I'm out of keys). ^O and ^W are particularly useful since they
delete whitespace. The one instance where 'd' is nice is with '%'. I'll
have to add a single key delete contents of bracketed expression in next
version of joe :-) Actually if you really feel this is important, make a macro.
It always amazes me how this tiny system seduces people into loving vi even
at the expense of it being modal. I would argue that vi is flawed even on
its own terms: if it allowed the cursor to go past the end of the line it
would not need all of those append equivalents of its insert commands (I
could never stand ed for this reason either). Incidentally, if you like
being able to fit commands together you should check out the old editor
mined- it even has conditional loops right at your fingertips.
>How about 'delete 5 words'?
^K \ 5 ^W- but for five words it's easier to hit ^W or ^O five times. In vi
you need 5dw because hitting dw five times is a pain.
>As for the other stuff I feel lacking there... well, equivalents
>of !<movement>command and :g/regex/<command>. ctags support (albeit I
>didn't use JOE for quite a while so I don't know whether it has it or not
>now). Multiple paste buffers. Again, it's all doable, but the basic command
>set is *way* too irregular and putting all that stuff there will lead to
>monster.
Joe has had block filtering and ctags for a long time. Joe doesn't need
multiple paste buffers- instead it has a stack for block marks and
bookmarks. I suppose :g can be nice, but the things for which it is usually
used are to overcome other shortcommings of vi- vi is limited to regexps in
its search strings (joe can match balanced C expressions with \c, limit
search & replace to the marked block, or limit search & replace to a
rectangular block), vi regexps can't cross lines (in joe you just use \n),
vi doesn't have column operations (joe can even filter a rectangular block
through a shell command). Joe is well integrated with unix- it can be used
as a filter (joe -), and every prompt has history and allows you to use !
(for executing a shell command), >> (for appending) and file,start,size (for
editing raw disks). Of course joe also has shell windows, cursor position
history, multiple windows, incremental search, math, recursive keyboard
macros with query waits, and undo/redo. Joe can edit MS-DOS files directly
and even has vi's numbered lines mode. I know many of these are addressed
in modern versions of vi (like the very excellent vim), but I don't feel
that joe should be so quickly dismissed as a good programmer's editor.
Anyway, I glad that you at least didn't complain that it lacks color syntax
highlighting.
--
/* [EMAIL PROTECTED] (192.74.137.5) */ /* Joseph H. Allen */
int a[1817];main(z,p,q,r){for(p=80;q+p-80;p-=2*a[p])for(z=9;z--;)q=3&(r=time(0)
+r*57)/7,q=q?q-1?q-2?1-p%79?-1:0:p%79-77?1:0:p<1659?79:0:p>158?-79:0,q?!a[p+q*2
]?a[p+=a[p+=q]=q]=q:0:0;for(;q++-1817;)printf(q%79?"%c":"%c\n"," #"[!a[q-1]]);}
------------------------------
From: Florian Erhard <[EMAIL PROTECTED]>
Subject: accessing a filesystem out of a module
Date: 30 Mar 1999 16:03:58 GMT
I'm creating a new filesystem as a kernel module. No
problem so far. My filesystem though doesn't use a
block device as underlying storage but another
filesystem. So I have to use open, close, read, write,
lseek, truncate, create, etc. from within my module.
It seems those system calls aren't exported symbols.
What can I do to access these operations? Would it
be a solution to use a user-level process as server? How
could I talk to it?
thanks for all hints,
Florian
------------------------------
From: "Darrel Davis" <[EMAIL PROTECTED]>
Subject: Re: JDBC driver and JDBC-ODBC bridge
Date: Tue, 30 Mar 1999 08:45:34 -0500
I am currently using the JDBC driver for MySql and have gotten
JDBC drivers for various DB's from Openlink, Informix and Oracle.
Exactly what are you trying to find?
-darrel
Chung-Shu Yang <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Is there any JDBC-ODBC bridge and/or free JDBC source code for Linux
> available?
>
> (Sorry for sending this message twice. Wrong identity last time....)
>
------------------------------
From: Aries <[EMAIL PROTECTED]>
Subject: Keyboard and Mouse
Date: Wed, 31 Mar 1999 19:04:28 GMT
Hello, I'd need to read in multiple keypresses at once and am not sure
how to do it in linux
In DOS it was easy, just write an interrupt handler which reads the scan
codes from the keyboard and process them as nesecary. In linux tho it
seems a lot more difficult. Im not sure which device to open or what to
expect to read. I can seem to find any information on it. I dont want to
use lin_getch() or something like that as I need a non-blocking way of
getting multiple keypresses (multiple keys pressed at once). It would
also be nice if this worked both in the term and under x-windows (id
likely have to check that my window was the current one).
Id also like to be able to read the mickeys moved by the mouse, again in
X and in the term.
Any help or pointer to code would be greatly appreciated.
thanks
phil
------------------------------
From: [EMAIL PROTECTED] (Donal K. Fellows)
Crossposted-To:
comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.help,comp.unix.programmer
Subject: Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC.
Date: 30 Mar 1999 13:50:55 GMT
In article <[EMAIL PROTECTED]>,
G. Sumner Hayes <[EMAIL PROTECTED]> wrote:
> The command "grep" comes from the ed commands Global REgexp Print,
> which is g/re/p, where re is whatever string you're looking for.
> vi has supported this (:g/foo/p) for as long as it's been around.
That just greps within a single loaded file. Emacs grep does
something much better and more wonderful, since it treats the output
of grep like the list of errors from a compiler, allowing you to step
backwards and forwards across a large multifile search. Very useful
when you're modifying a useful global function to take an extra pair
of parameters...
Donal.
--
Donal K. Fellows http://www.cs.man.ac.uk/~fellowsd/ [EMAIL PROTECTED]
Department of Computer Science, University of Manchester, U.K. +44-161-275-6137
--
"And remember, evidence is nothing." - Stacy Strock <[EMAIL PROTECTED]>
------------------------------
From: "Paul F. Dietz" <[EMAIL PROTECTED]>
Crossposted-To: comp.unix.programmer
Subject: Re: [ANN] CodeWarrior for Red Hat Linux
Date: Tue, 30 Mar 1999 07:14:20 -0600
Sid Boyce wrote:
> This individual, like so many other authors of
> late have hung on the fragmentation of Linux as did a Mr. Gates
> MicroSomething who said some garbage about the incompatibility problems
> of Linux distributions, as an aside, it's the first time I've ever read
> (or heard) anything from Mr. Gates that was a coherent sentence, I
> always understand all the words he uses, but the meaning is very tricky,
> it usually ends with me asking "What did he say ?", then getting on with
> life.
Those who use run-on sentences should not throw stones.
Paul
------------------------------
Crossposted-To:
comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.help,comp.unix.programmer
Subject: Re: CodeWarror for Linux (was: Re: Programming tools for ...)
From: Klaus Schilling <[EMAIL PROTECTED]>
Date: 30 Mar 1999 15:26:22 +0200
[EMAIL PROTECTED] (Charlie Stross) writes:
> This has not yet shown up in MetroWerks or Red Hat PR material so
> it should be treated as a rumour -- but MetroWerks announced that
> they were working on a Linux port last year, and the substance of
> this rumour makes eminent sense for both companies.
Loyal GNU/Linux Users don't need and won't use propritary rubbish like
Codewarrior, but stick with classical GNU Tools which are the best possible
way to use a computer in a decent, righteous way.
Klaus Schilling
--
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Subject: Re: clone() or PThreads ???
Date: Wed, 31 Mar 1999 14:57:42 -0500
[It sounds like you really need to pick up a copy of Stevens'
Unix programming book]
Udo Giacomozzi wrote:
>
> First, thank you for your interesting reply!
> It seems I need to use fork() because the RTL is not reentrant. Ok.
>
> The reason why I want to share the memory is because my application is
> made completely object-oriented, including sound routines. The class
> functions and data should be accessible from both processes. I don't
> think this will be a problem because class functions are reentrant. I
> don't know if I really need shared memory. All I need is access to the
> classes (owned by the parent). So if I can pass the pointer to the
> class to the child and use it, there is no problem (no shared memory
> is needed). But I actually don't know if this is possible under Linux
> (as memory management is different than under Windows).
Hmm. This is going to be difficult if you truly need real-time
behaviour. You cannot pass a pointer to the classes between two
processes -- if you fork() then the processes have separate address
spaces.
Think of fork() as being akin to spawn() under Windows, but
it spawn()s another copy of the current process that has nearly
identical state to the current process -- anything that changes in the
parent process after the fork() is invisible to the child and
vice-versa. fork() (or vfork()) followed by exec() is equal to
a real spawn() call in 95/NT (starting a new process at the beginning
of main()). AFAIK, Win32 has protected memory, so you couldn't pass
pointers between spawn()'d processes and have it work there. It
certainly won't work under Linux or other modern OSes.
So you need some form of inter-process communication (IPC). Under
Linux, that means some combination of:
1. signals (either normal or queued)
2. SysV IPC (shared memory, semaphores)
3. mmap() (shared memory)
4. fcntl locking (for synchronization)
5. sockets/pipes
(there are other solutions, too; those are just the most common)
You could use shared memory (SysV or mmap, mmap is slightly nicer),
but you'll be left with a couple of problems:
1. You cannot current lock shared memory into RAM on Linux. This
basically kills shm for real-time applications -- if the shm is
swapped out, your "real-time" process will have to wait for it to
be swapped back in from disk. Bad mojo.
2. You need to make sure that all of your class objects are allocated
in the shared memory region. Probably pretty straightforward.
3. You say that your class hierarchy is reentrant -- that is absolutely
critical when accessing data in shared memory. Good, this one's not
a problem.
It sounds like #1 is the real bummer. Moreover, since your class
hierarchy is reentrant it seems like you actually would prefer a
multithreaded-solution. Unfortunately, you're stuck with a crufty old
run-time, so you're going to have to kludge -- either work with shm
and dump enough RAM into the machine that you don't start swapping out
the shm regions, switch to a message-passing API and use sockets, or
figure out some way of isolating all of the calls into the run-time and
make them from one thread or put locks around them.
Using shm (despite the lack of mlock for shm) is actually not a bad
idea if you don't need absolute real-time. If you have enough RAM,
there shouldn't be a problem. If failure to have real-time behaviour
is going to be mission-critical, you ought to be looking at rtlinux
or psOS or some other hard real-time OS anyway.
The best solution would be to get a reentrant run-time and just use
threads. Is there any way you can do that? Do you have the source
code for the runtime?
--Sumner
------------------------------
** 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
******************************
