Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.c > #12450 > unrolled thread

undefined reference to `link'

Back to article view | Back to comp.lang.c

Contents

  undefined reference to `link' Jinsong Zhao <jszhao@yeah.net> - 2011-11-01 23:41 +0800
    Re: undefined reference to `link' Mark Bluemel <mark_bluemel@pobox.com> - 2011-11-01 15:46 +0000
      Re: undefined reference to `link' Jinsong Zhao <jszhao@yeah.net> - 2011-11-02 00:01 +0800
        Re: undefined reference to `link' Mark Bluemel <mark_bluemel@pobox.com> - 2011-11-01 16:21 +0000
          Re: undefined reference to `link' jgharston <jgh@arcade.demon.co.uk> - 2011-11-02 04:43 -0700
            Re: undefined reference to `link' Keith Thompson <kst-u@mib.org> - 2011-11-02 14:10 -0700
              Re: undefined reference to `link' jgharston <jgh@arcade.demon.co.uk> - 2011-11-03 17:08 -0700
                Re: undefined reference to `link' Keith Thompson <kst-u@mib.org> - 2011-11-03 17:17 -0700
                  Re: undefined reference to `link' gordonb.hsqhk@burditt.org (Gordon Burditt) - 2011-11-03 23:11 -0500
            Re: undefined reference to `link' gordonb.d5wxg@burditt.org (Gordon Burditt) - 2011-11-02 16:28 -0500
              Re: undefined reference to `link' Keith Thompson <kst-u@mib.org> - 2011-11-02 15:30 -0700
                Re: undefined reference to `link' gordonb.wk71i@burditt.org (Gordon Burditt) - 2011-11-02 19:33 -0500
                  Re: undefined reference to `link' Keith Thompson <kst-u@mib.org> - 2011-11-02 19:24 -0700
        Re: undefined reference to `link' Acid Washed China Blue Jeans <chine.bleu@yahoo.com> - 2011-11-01 10:02 -0700
          Re: undefined reference to `link' gordonb.ztlqg@burditt.org (Gordon Burditt) - 2011-11-02 19:46 -0500
            Re: undefined reference to `link' Seebs <usenet-nospam@seebs.net> - 2011-11-03 18:35 +0000
              Re: undefined reference to `link' Kaz Kylheku <kaz@kylheku.com> - 2011-11-03 19:56 +0000
                Re: undefined reference to `link' Keith Thompson <kst-u@mib.org> - 2011-11-03 14:04 -0700
                Re: undefined reference to `link' Seebs <usenet-nospam@seebs.net> - 2011-11-03 21:45 +0000
                Re: undefined reference to `link' Acid Washed China Blue Jeans <chine.bleu@yahoo.com> - 2011-11-03 15:53 -0700
        Re: undefined reference to `link' Mike Manilone <zhtx10@gmail.com> - 2011-11-02 04:42 +0000
        Re: undefined reference to `link' Geoff <geoff@invalid.invalid> - 2011-11-03 14:52 -0700
          Re: undefined reference to `link' Jinsong Zhao <jszhao@yeah.net> - 2011-11-04 08:19 +0800
            Re: undefined reference to `link' Geoff <geoff@invalid.invalid> - 2011-11-03 21:32 -0700
              Re: undefined reference to `link' Jinsong Zhao <jszhao@yeah.net> - 2011-11-04 13:25 +0800
              Re: undefined reference to `link' Jinsong Zhao <jszhao@yeah.net> - 2011-11-04 13:28 +0800
                Re: undefined reference to `link' Geoff <geoff@invalid.invalid> - 2011-11-03 23:01 -0700
    Re: undefined reference to `link' Kaz Kylheku <kaz@kylheku.com> - 2011-11-03 02:40 +0000

Page 1 of 2  [1] 2  Next page →

#12450 — undefined reference to `link'

Hi there,

I tried to compile a c program 
(http://people.sc.fsu.edu/~jburkardt/c_src/f77split/f77split.c) using 
MinGW/gcc 4.6.1:

gcc f77split.c

however, I get the following error message:

f77split.c:(.text+0x4a9): undefined reference to `link'
collect2: ld returned 1 exit status

However, compiled the same code with gcc 4.4.x on a Linux machine, and 
succeed.

Does it mean I missing something on my Windows XP with MinGW/gcc 4.6.1?

Any suggestions will be really appreciated.

Regards,
Jinsong

[toc] | [next] | [standalone]

#12452

On 11/01/2011 03:41 PM, Jinsong Zhao wrote:
> Hi there,
>
> I tried to compile a c program
> (http://people.sc.fsu.edu/~jburkardt/c_src/f77split/f77split.c) using
> MinGW/gcc 4.6.1:
>
> gcc f77split.c
>
> however, I get the following error message:
>
> f77split.c:(.text+0x4a9): undefined reference to `link'
> collect2: ld returned 1 exit status
>
> However, compiled the same code with gcc 4.4.x on a Linux machine, and
> succeed.
>
> Does it mean I missing something on my Windows XP with MinGW/gcc 4.6.1?
>
> Any suggestions will be really appreciated.

The code tries to use the link() (and unlink() I think) routines which 
are part of POSIX, not C. They'll be present on a Linux/Unix system, but 
probably not on Windows, even wth MinGW.

[toc] | [prev] | [next] | [standalone]

#12455

On 2011-11-1 23:46, Mark Bluemel wrote:
> On 11/01/2011 03:41 PM, Jinsong Zhao wrote:
>> Hi there,
>>
>> I tried to compile a c program
>> (http://people.sc.fsu.edu/~jburkardt/c_src/f77split/f77split.c) using
>> MinGW/gcc 4.6.1:
>>
>> gcc f77split.c
>>
>> however, I get the following error message:
>>
>> f77split.c:(.text+0x4a9): undefined reference to `link'
>> collect2: ld returned 1 exit status
>>
>> However, compiled the same code with gcc 4.4.x on a Linux machine, and
>> succeed.
>>
>> Does it mean I missing something on my Windows XP with MinGW/gcc 4.6.1?
>>
>> Any suggestions will be really appreciated.
>
> The code tries to use the link() (and unlink() I think) routines which
> are part of POSIX, not C. They'll be present on a Linux/Unix system, but
> probably not on Windows, even wth MinGW.

yes, the code use link and unlink routines.

Is it possible to compile the code on Windows? If yes, how to change the 
code?

Thanks.

Regards,
Jinsong

[toc] | [prev] | [next] | [standalone]

#12457

On 11/01/2011 04:01 PM, Jinsong Zhao wrote:
> On 2011-11-1 23:46, Mark Bluemel wrote:
>> On 11/01/2011 03:41 PM, Jinsong Zhao wrote:
>>> Hi there,
>>>
>>> I tried to compile a c program
>>> (http://people.sc.fsu.edu/~jburkardt/c_src/f77split/f77split.c) using
>>> MinGW/gcc 4.6.1:
>>>
>>> gcc f77split.c
>>>
>>> however, I get the following error message:
>>>
>>> f77split.c:(.text+0x4a9): undefined reference to `link'
>>> collect2: ld returned 1 exit status
>>>
>>> However, compiled the same code with gcc 4.4.x on a Linux machine, and
>>> succeed.
>>>
>>> Does it mean I missing something on my Windows XP with MinGW/gcc 4.6.1?
>>>
>>> Any suggestions will be really appreciated.
>>
>> The code tries to use the link() (and unlink() I think) routines which
>> are part of POSIX, not C. They'll be present on a Linux/Unix system, but
>> probably not on Windows, even wth MinGW.
>
> yes, the code use link and unlink routines.
>
> Is it possible to compile the code on Windows? If yes, how to change the
> code?

You need to work out what the link and unlink routines are being used 
for - I think it's to rename a temporary work file to become an output
file - and then write appropriate code to do that using routines which 
are available on Windows. You probably should ask in a windows forum for 
help with those routines.

[toc] | [prev] | [next] | [standalone]

#12528

Mark Bluemel wrote:
> > yes, the code use link and unlink routines.
> You need to work out what the link and unlink routines are being used
> for - I think it's to rename a temporary work file to become an output

If you're using link/unlink to rename a file, the stdio library
contains
the standard rename() function.

Always start by looking in your standard libraries for the
functionality
you're looking for before going trekking off in other directions.

JGH

[toc] | [prev] | [next] | [standalone]

#12551

jgharston <jgh@arcade.demon.co.uk> writes:
> Mark Bluemel wrote:
>> > yes, the code use link and unlink routines.
>> You need to work out what the link and unlink routines are being used
>> for - I think it's to rename a temporary work file to become an output
>
> If you're using link/unlink to rename a file, the stdio library
> contains
> the standard rename() function.
>
> Always start by looking in your standard libraries for the
> functionality you're looking for before going trekking off in other
> directions.

rename() doesn't do what the Unix link() system call does.

link() creates a new name for a file.  The new name and the old name
both refer to the same file in the same way; neither of them is, in any
real sense, the original.

The C standard library doesn't provide anything with equivalent
functionality -- probably because there's no guarantee that the
underlying OS or filesystem is capable of supporting it.

(I tried using the "ln" command under Cygwin, a Unix emulation layer
that runs on top of MS Windows, and got some odd and inconsistent
results.)

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"

[toc] | [prev] | [next] | [standalone]

#12644

Keith Thompson wrote:
> jgharston wrote:
> > If you're using link/unlink to rename a file, the stdio library
> > contains the standard rename() function.
>
> rename() doesn't do what the Unix link() system call does.

That's why I wrote "If you're using link/unlink to rename a file..."

JGH

[toc] | [prev] | [next] | [standalone]

#12646

jgharston <jgh@arcade.demon.co.uk> writes:
> Keith Thompson wrote:
>> jgharston wrote:
>> > If you're using link/unlink to rename a file, the stdio library
>> > contains the standard rename() function.
>>
>> rename() doesn't do what the Unix link() system call does.
>
> That's why I wrote "If you're using link/unlink to rename a file..."

Ok, I didn't pay enough attention to the context.

But why would you use link and unlink to rename a file in the first
place?  rename() exists both as a Unix system call and as a standard C
library function.

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"

[toc] | [prev] | [next] | [standalone]

#12657

You might use link/unlink to rename a file on UNIX because it's
archaic code.  On UNIX v7, remove() and rename() didn't exist, and
the link/unlink dance was the only way to rename a file.  Neither
remove() nor rename() existed in K&R 1 (published 1978).  They did
by the time of C89, and probably were added in the mid-1980's.  I
don't think they existed in 1983 in the Microsoft Xenix code (which
was derived from real UNIX code).

In newly-written code, there's no excuse for it.

[toc] | [prev] | [next] | [standalone]

#12555

> If you're using link/unlink to rename a file, the stdio library
> contains
> the standard rename() function.
> 
> Always start by looking in your standard libraries for the
> functionality
> you're looking for before going trekking off in other directions.

Note that the link/unlink procedure assumes that this can be done
repeatedly:  if the destination file already exists (left over from
the last time the code was executed), it is removed; then the rename
is done.

The standard rename() function does not necessarily do this.  C
says that if the destination name already exists, the result is
implementation-defined.  For POSIX (these systems would have link()
/ unlink()), the file is renamed.  For Windows, the rename() fails.

Use of the C standard functions remove(), then rename() will restore
the behavior to what I believe was intended, except that the POSIX
guarantee that the destination file always exists is no longer
valid.

The original code and replacement remove()/rename() code both have
potential race conditions if other programs might be creating or
deleting files with the source or destination names while the code
is running.

[toc] | [prev] | [next] | [standalone]

#12561

gordonb.d5wxg@burditt.org (Gordon Burditt) writes:
[...]
> Note that the link/unlink procedure assumes that this can be done
> repeatedly:  if the destination file already exists (left over from
> the last time the code was executed), it is removed; then the rename
> is done.

Not according to the link(2) man page on my system:

    If newpath exists it will not be overwritten.

And link() is *not* a form of renaming.  It creates a new name for a
file; the old name continues to exist.

[...]

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"

[toc] | [prev] | [next] | [standalone]

#12564

Read the code containing two unlink() calls and one link() call.
*THAT CODE*, not link() alone, performs a rough equivalent of a
rename().  I don't care about the behavior of link() if the target
already exists, except in race conditions, because I just deleted
the target.  Before the invention of rename() and remove(), this
sequence of calls was how you did a file rename on early UNIX:

	unlink(to);		/* failure is expected if file doesn't exist */
	link(from, to);		/* if this fails and you don't check errors */
				/* you'll lose the file */
	unlink(from);
instead of:
	rename(from, to);

Error checking is not shown but is pretty mandatory or you'll lose the
file under the wrong circumstances (like attempting a cross-filesystem
rename).  

There are, of course, some exceptions to the rough equivalence:

The unlink/link code has guaranteed behavior if the "to" file already
exists; the equivalent rename() does not unless you are using POSIX rename
rather than C Standard rename().

There are race conditions if someone else creates or deletes the "to" or
"from" files during the execution of the sequence.

Both rename() and link() are allowed to fail if "to" and "from" aren't on
the same filesystem, but if the files are in the same directory, you're
fairly (but not absolutely) safe.

[toc] | [prev] | [next] | [standalone]

#12568

gordonb.wk71i@burditt.org (Gordon Burditt) writes:
> Read the code containing two unlink() calls and one link() call.
> *THAT CODE*, not link() alone, performs a rough equivalent of a
> rename().
[...]

What is this in response to?  (check parent article) Oh, I see, this is
a followup to my post, and since I've told you not to quote my words
without attribution, you've decided to post followups without quoting
me at all.

In the future, I request that you don't do this.

I explicitly deny permission to quote my words without attribution.  I
ask that you not post followups to my articles without quoting my words.

The obvious solution is to quote with attribution, as everyone else
does.  If you're paranoid, you can use an attribution line that doesn't
claim that I *really* wrote the quoted words, like "Someone claiming to
be Keith Thompson writes:".

Another solution would be just not to post followups to anything I
write.  And if that's inconvenient, I suggest that it's entirely your
own fault.

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"

[toc] | [prev] | [next] | [standalone]

#12463

In article <j8p55i$v4p$1@dont-email.me>, Jinsong Zhao <jszhao@yeah.net> wrote:

> On 2011-11-1 23:46, Mark Bluemel wrote:
> > On 11/01/2011 03:41 PM, Jinsong Zhao wrote:
> >> Hi there,
> >>
> >> I tried to compile a c program
> >> (http://people.sc.fsu.edu/~jburkardt/c_src/f77split/f77split.c) using
> >> MinGW/gcc 4.6.1:
> >>
> >> gcc f77split.c
> >>
> >> however, I get the following error message:
> >>
> >> f77split.c:(.text+0x4a9): undefined reference to `link'
> >> collect2: ld returned 1 exit status
> >>
> >> However, compiled the same code with gcc 4.4.x on a Linux machine, and
> >> succeed.
> >>
> >> Does it mean I missing something on my Windows XP with MinGW/gcc 4.6.1?
> >>
> >> Any suggestions will be really appreciated.
> >
> > The code tries to use the link() (and unlink() I think) routines which
> > are part of POSIX, not C. They'll be present on a Linux/Unix system, but
> > probably not on Windows, even wth MinGW.
> 
> yes, the code use link and unlink routines.
> 
> Is it possible to compile the code on Windows? If yes, how to change the 
> code?

I think there are adaption libraries which convert Posix calls to Windows calls. 
Cygwin?

-- 
White folks think they're at the top,  |   Faded, patched, secret so tight...
ask any proud white male.              |       I'm whoever you want me to be.
A million years of evolution,          |  Annoying Usenet one post at a time.
and we get Danny Quayle.               |    At least I can stay in character.

[toc] | [prev] | [next] | [standalone]

#12565

> I think there are adaption libraries which convert Posix calls to Windows calls. 
> Cygwin?

In this case, it would be difficult.  If you are renaming a file
by link() followed by unlink(), and trying to make this work on
Windows, and the program was terminated after the link(), this
requires a state of the filesystem I don't think Windows supports
(at least not on a FAT filesystem, and I'm not sure about NTFS).
(and file system checking would attempt to "fix" it.)

It might work in this case if link() copied the file, and the file
permissions.  There are other uses of link() where that won't work,
such as the expectation that changes made to one, uh, "copy" appear
in both, since there aren't supposed to be two separate copies.

[toc] | [prev] | [next] | [standalone]

#12611

On 2011-11-03, Gordon Burditt <gordonb.ztlqg@burditt.org> wrote:
>> [unattributed quote]

This is really annoying.  It is completely pointless.  It does not
provide you with any legal protection that I've ever heard of, and it
makes it much harder to follow your "contributions" to conversations.

... enough harder that I, for one, am just plonking you because it's
not worth the hassle of trying to figure out who you don't think you
were really responding to.

-s
-- 
Copyright 2011, all wrongs reversed.  Peter Seebach / usenet-nospam@seebs.net
http://www.seebs.net/log/ <-- lawsuits, religion, and funny pictures
http://en.wikipedia.org/wiki/Fair_Game_(Scientology) <-- get educated!
I am not speaking for my employer, although they do rent some of my opinions.

[toc] | [prev] | [next] | [standalone]

#12617

On 2011-11-03, Seebs <usenet-nospam@seebs.net> wrote:
> On 2011-11-03, Gordon Burditt <gordonb.ztlqg@burditt.org> wrote:
>>> [unattributed quote]
>
> This is really annoying.  It is completely pointless.  It does not

I don't see the big problem. Gordon left the parent article reference intact,
so it's easy to see where the quote came from.

In slrn: ESC-P fetches the parent article. There it is: "Acid Washed".

[toc] | [prev] | [next] | [standalone]

#12625

Kaz Kylheku <kaz@kylheku.com> writes:
> On 2011-11-03, Seebs <usenet-nospam@seebs.net> wrote:
>> On 2011-11-03, Gordon Burditt <gordonb.ztlqg@burditt.org> wrote:
>>>> [unattributed quote]
>>
>> This is really annoying.  It is completely pointless.  It does not
>
> I don't see the big problem. Gordon left the parent article reference intact,
> so it's easy to see where the quote came from.
>
> In slrn: ESC-P fetches the parent article. There it is: "Acid Washed".

Then why did you bother to quote the parent article?

In the newsreader I use (Gnus), it's easy to jump to the parent
article, but it can be difficult to get back (or maybe there's a
command I don't know about).  The vast majority of posters here quote
enough of the parent article that I rarely need to bother.  I find
Gordon's stubborn refusal to attribute quoted text inexplicable
and annoying.  (Though I do give him credit for respecting my wishes
by not quoting my text without attribution.)

And there's a solution that would simultaneously make Gordon's
articles as easy to read as anyone else's and satisfy his paranoia.

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"

[toc] | [prev] | [next] | [standalone]

#12628

On 2011-11-03, Kaz Kylheku <kaz@kylheku.com> wrote:
> I don't see the big problem. Gordon left the parent article reference intact,
> so it's easy to see where the quote came from.

If the other article happens to be available.

> In slrn: ESC-P fetches the parent article. There it is: "Acid Washed".

Some people read news offline.  I'm one of them.  And while in theory I can
go around navigating article trees to look stuff up, in practice, it breaks
the general rule that I don't like to spend more effort than the reward I
get.

I don't desperately need to read his posts.  I'm not going to if making sense
of them requires more effort than making sense of normal Usenet posts does.

-s
-- 
Copyright 2011, all wrongs reversed.  Peter Seebach / usenet-nospam@seebs.net
http://www.seebs.net/log/ <-- lawsuits, religion, and funny pictures
http://en.wikipedia.org/wiki/Fair_Game_(Scientology) <-- get educated!
I am not speaking for my employer, although they do rent some of my opinions.

[toc] | [prev] | [next] | [standalone]

#12638

In article <20111103155231.299@kylheku.com>, Kaz Kylheku <kaz@kylheku.com> 
wrote:

> On 2011-11-03, Seebs <usenet-nospam@seebs.net> wrote:
> > On 2011-11-03, Gordon Burditt <gordonb.ztlqg@burditt.org> wrote:
> >>> [unattributed quote]
> >
> > This is really annoying.  It is completely pointless.  It does not
> 
> I don't see the big problem. Gordon left the parent article reference intact,
> so it's easy to see where the quote came from.
> 
> In slrn: ESC-P fetches the parent article. There it is: "Acid Washed".

MT-Newswatcher has Special >> Open All References

It opens all the cumulative References: still on the server.

-- 
White folks think they're at the top,  |   Faded, patched, secret so tight...
ask any proud white male.              |       I'm whoever you want me to be.
A million years of evolution,          |  Annoying Usenet one post at a time.
and we get Danny Quayle.               |    At least I can stay in character.

[toc] | [prev] | [next] | [standalone]

Page 1 of 2  [1] 2  Next page →

Back to top | Article view | comp.lang.c

csiph-web