Re: Assignment make pointers

Re: Assignment make pointers

[r_zaca]

[-- Attachment #1: Mail message body --]
[-- Type: text/plain, Size: 1063 bytes --]

  If you take a look in "man 3 basename", you'll see that basename function 
just returns a pointer to a char. In the case above you need to "cast" the 
pointer returned to be of type int. 
  Like: prog_ptr = (int *) basename (argv[0]); 
  I hope it will help you. 
  Good luck. 


Hello, 

New to the list and to C. Read O'Reilly's Practical C Programming and 
another C book and am now trying to evolve beyond helloworld.c. When I 
compile the snippet below, gcc-3.4.2 on Fedora Core 3 spits out: 
test2.c:11: warning: assignment makes pointer from integer without a 
cast 
test2.c:12: warning: passing arg 1 of `usage' from incompatible pointer 
type 

I just can't figure out why it does that and how to fix it since my 
pointer/casting/C knowledge is definitely lacking. The snippet does work 
ok. Anyone care to enlighten me? 

#include <stdio.h> 

void usage(int *prog_ptr) 
{ 
	fprintf(stdout,"progname is: %s\n",*prog_ptr); 
} 
	 
int main(int argc, char *argv[]) 
{ 
	int *prog_ptr; 
	prog_ptr = basename (argv[0]); 
	usage(&prog_ptr); 
	return(0); 
} 

Re: Assignment make pointers

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 1322 bytes --]

On Mon, 2004-12-13 08:04:32 -0200, r_zaca <r_zaca@ig.com.br>
wrote in message <20041213_100432_025782.r_zaca@ig.com.br>:
Content-Description: Mail message body
>   If you take a look in "man 3 basename", you'll see that basename function 
> just returns a pointer to a char. In the case above you need to "cast" the 
> pointer returned to be of type int. 
>   Like: prog_ptr = (int *) basename (argv[0]); 

Don't do that. Usually, you can write even quite large programs without
using casts at all. Most of the time, a cast points to spots in your
programs that are badly designed from the type of your variables point
of view. If your variables are of senseful types and obey a well-thought
structure (like using enums and unions instead of #defines and casts),
you usually don't need to cast.

Also, I usually suggest to compile code with "-Wall" -- that'll show you
probably lots of odd constructs that should be written differently (or
using proper types).

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Assignment make pointers

[Eric Bambach]

On Monday 13 December 2004 04:10 am, Jan-Benedict Glaw wrote:
> On Mon, 2004-12-13 08:04:32 -0200, r_zaca <r_zaca@ig.com.br>
> wrote in message <20041213_100432_025782.r_zaca@ig.com.br>:
> Content-Description: Mail message body
>
> >   If you take a look in "man 3 basename", you'll see that basename
> > function just returns a pointer to a char. In the case above you need to
> > "cast" the pointer returned to be of type int.
> >   Like: prog_ptr = (int *) basename (argv[0]);
>
> Don't do that. Usually, you can write even quite large programs without
> using casts at all. Most of the time, a cast points to spots in your
> programs that are badly designed from the type of your variables point
> of view. If your variables are of senseful types and obey a well-thought
> structure (like using enums and unions instead of #defines and casts),
> you usually don't need to cast.
> Also, I usually suggest to compile code with "-Wall" -- that'll show you
> probably lots of odd constructs that should be written differently (or
> using proper types).

True enough. The problem intrigued me enough that I test compiled with -Wall 
and the compiler complains of an implicit declaration of basename. It seems 
he was missing #include <libgen.h>. Adding this cleared everything up.

What escapes me though is how the compiler could compile without knowing the 
definition. Isnt it REALLY presumptuous and dangerous for the compiler to 
assume function definitions? I assume basename is in the standard c library 
though or it certainly wouldnt link. Can anyone with more experience 
elaborate?

> MfG, JBG

-- 

-EB

Re: Assignment make pointers

[Eric Bambach]

On Monday 13 December 2004 08:51 am, Eric Bambach wrote:
> What escapes me though is how the compiler could compile without knowing
> the definition. Isnt it REALLY presumptuous and dangerous for the compiler
> to assume function definitions? I assume basename is in the standard c
> library though or it certainly wouldnt link. Can anyone with more
> experience elaborate?

Sorry to reply to my own post, but nevermind. Its too early (someone called 
and woke me up :) and I've been programming in C++ too long where prototypes 
are required. gcc will compile happily without libgen.h while g++ won't. That 
answers my own question. 

Sorry for the noise.

-- 

-EB

C Newbie

[Darren Sessions]

I need to know (before I get flamed) if this is an appropriate spot to post
questions on C programming or not and the very most newbie level. I'm
attempting a transition from Perl and, while I understand a lot of the
fundamentals of C, am having a hard time with string (char) manipulation.

Thanks,

 - Darren

Re: Assignment make pointers

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 772 bytes --]

On Mon, 2004-12-13 08:51:44 -0600, Eric Bambach <eric@cisu.net>
wrote in message <200412130851.44586.eric@cisu.net>:

> What escapes me though is how the compiler could compile without knowing the 
> definition. Isnt it REALLY presumptuous and dangerous for the compiler to 

The "type defaults to int" thing. This is why the initial writer got an
error message that an int type is about to be converted to pointer type.

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: C Newbie

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 1008 bytes --]

On Mon, 2004-12-13 10:01:28 -0500, Darren Sessions <dsessions@ionosphere.net>
wrote in message <BDE318F8.1618%dsessions@ionosphere.net>:
> I need to know (before I get flamed) if this is an appropriate spot to post
> questions on C programming or not and the very most newbie level. I'm
> attempting a transition from Perl and, while I understand a lot of the
> fundamentals of C, am having a hard time with string (char) manipulation.

Start asking. However, start a question with a new email, not by
pressing you email client's "reply" button and deleting any old text.
Hijacking threads is bad habit and makes proper email sorting a hard
challange...

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: C Newbie

[Eric Bambach]

On Monday 13 December 2004 09:01 am, you wrote:
> I need to know (before I get flamed) if this is an appropriate spot to post
> questions on C programming or not and the very most newbie level. I'm
> attempting a transition from Perl and, while I understand a lot of the
> fundamentals of C, am having a hard time with string (char) manipulation.
>
> Thanks,
>
>  - Darren

As JBG advised yes please don't hijack threads. It confuses me AND my mailer. 
Other than that I would be happy to assist on the list if I can, even though 
my programming skills aren't great yet. My advice is to limit your questions 
though. Try to do most of the reasearch off list and ask specifc questions 
when something stumps you or doesnt work as expected after you've tried to 
debug it yourself. If it looks like you didn't try (ex. throwing code at us 
with the subject it doesnt compile) or cant give us specific code with 
questions to help on you're less likely to get any useful information off the 
list.

Other than that, I hope I can be of help and I hope the list gurus agree with 
me.

-- 

-EB

Defunct Processes

[Darren Sessions]

The application I'm working on right now gets called from another
application. When my program is finished, I just use exit(0); to quit, but
if I do a ps there are a ton of [applicationname] <defunct> messages.

I'm using Suse Enterprise Linux 9.0 with the 2.6 Kernel.

Thanks,

 - Darren

Re: Defunct Processes

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 1425 bytes --]

On Tue, 2004-12-28 21:13:56 -0500, Darren Sessions <dsessions@ionosphere.net>
wrote in message <BDF77D14.1AE7%dsessions@ionosphere.net>:
> The application I'm working on right now gets called from another
> application. When my program is finished, I just use exit(0); to quit, but
> if I do a ps there are a ton of [applicationname] <defunct> messages.

This is a fault of the calling application. Your program became a
zombie. That's the remaining rest of a process that already finished.
(Basically, it consists of accounting information, the process' exit
value and probably of the signal number that led to the process' death).

This information needs to be handled (the parent process needs to wait()
or wait4() it, or using other methods). Normally, the parent process
should handle the SIGCHLD signal; it's generated upon a child's death.
This would allow the parent to call wait() as long as there are finished
childs (remember, you only know the signal got fired, but you don't know
how often :)  Calling wait() then will finally bury the finished child.

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Defunct Processes

[Daniel Souza]

you can

signal(SIGCHLD, SIG_IGN);

to ignore childs termination notifications, so, the parent dont need
to wait() for childs, they just go without any parent notification,
and no zombies.