GNU bug report logs - #26044
coreutils-8.27 fails to build

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 26044 in the body.
You can then email your comments to 26044 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: J Rogowsky <geometriker <at> gmail.com>
To: bug-coreutils <at> gnu.org
Subject: coreutils-8.27  fails to build
Date: Thu, 9 Mar 2017 23:08:58 -0700

coreutils-8.27  fails to build.

function renameat, found in src/force-link.c, is not found on my  
system, which is an iBook G4 running OSX 10.5.8.
Therefore the link step fails, citing the linker cannot find _renameat.


I see there is a similar function in lib/at-func2.c which could be  
used, but I chose to solve this problem in the simplest way possible  
by defining renameat in src/force-link.c like so:

diff src/force-link.c src/force-link.c.OEM
86,93d85
< /* see: https://www.freebsd.org/cgi/man.cgi?query=rename&sektion=2  */
< static int
< renameat(int fromfd,const char *from,int tofd,const char *to)
< {
< 	return rename(from,to);
< }
<
<

Perhaps there is a better solution?
Such as using at_func2 in src/force-link.c  like so:

static int
renameat(int fromfd,const char *from,int tofd,const char *to)
{
	return at_func2(fromfd,from,tofd,to,rename);
}

????????????????

BTW, I had only one test FAIL in "make check":  tests/misc/sort-debug- 
keys
Is this related to the above?


If you would like further assistance on this matter, just let me know  
how I could help.

Thanks,
J Rogowsky

Message #8 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: J Rogowsky <geometriker <at> gmail.com>, 26044 <at> debbugs.gnu.org
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Thu, 9 Mar 2017 23:22:56 -0800

On 09/03/17 22:08, J Rogowsky wrote:
> coreutils-8.27  fails to build.
> 
> function renameat, found in src/force-link.c, is not found on my  
> system, which is an iBook G4 running OSX 10.5.8.
> Therefore the link step fails, citing the linker cannot find _renameat.


Drats. It looks like the coreutils source repo will need something like:

diff --git a/bootstrap.conf b/bootstrap.conf
index 7def1f9..30ce621 100644
--- a/bootstrap.conf
+++ b/bootstrap.conf
@@ -202,6 +202,7 @@ gnulib_modules="
   regex
   remove
   rename
+  renameat
   rmdir
   root-dev-ino
   rpmatch


> I see there is a similar function in lib/at-func2.c which could be  
> used, but I chose to solve this problem in the simplest way possible  
> by defining renameat in src/force-link.c like so:
> 
> diff src/force-link.c src/force-link.c.OEM
> 86,93d85
> < /* see: https://www.freebsd.org/cgi/man.cgi?query=rename&sektion=2  */
> < static int
> < renameat(int fromfd,const char *from,int tofd,const char *to)
> < {
> < 	return rename(from,to);
> < }

I've not analysed whether that will work (in all cases).

> Perhaps there is a better solution?
> Such as using at_func2 in src/force-link.c  like so:
> 
> static int
> renameat(int fromfd,const char *from,int tofd,const char *to)
> {
> 	return at_func2(fromfd,from,tofd,to,rename);
> }
> 
> ????????????????

I've applied that locally and generated an untested tarball with `make dist`:
https://www.pixelbeat.org/cu/coreutils-8.27.4-3c9d7.tar.xz

Could you test that one out?

> BTW, I had only one test FAIL in "make check":  tests/misc/sort-debug- 
> keys
> Is this related to the above?

I'm guessing that "failure" might be locale related.
Could you attach the tests/test-suite.log file generated with:
  make TESTS=tests/misc/sort-debug-keys.sh SUBDIRS=. VERBOSE=yes check

thanks,
Pádraig

Message #11 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Pádraig Brady <P <at> draigBrady.com>,
 J Rogowsky <geometriker <at> gmail.com>, 26044 <at> debbugs.gnu.org
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Thu, 9 Mar 2017 23:33:06 -0800

Pádraig Brady wrote:

> Drats. It looks like the coreutils source repo will need something like:
>
> diff --git a/bootstrap.conf b/bootstrap.conf
> index 7def1f9..30ce621 100644
> --- a/bootstrap.conf
> +++ b/bootstrap.conf
> @@ -202,6 +202,7 @@ gnulib_modules="
>    regex
>    remove
>    rename
> +  renameat
>    rmdir
>    root-dev-ino
>    rpmatch

Yes, that looks like the right fix. Sorry about that.


>> diff src/force-link.c src/force-link.c.OEM
>> 86,93d85
>> < /* see: https://www.freebsd.org/cgi/man.cgi?query=rename&sektion=2  */
>> < static int
>> < renameat(int fromfd,const char *from,int tofd,const char *to)
>> < {
>> < 	return rename(from,to);
>> < }
>
> I've not analysed whether that will work (in all cases).

It should be an adequate workaround for coreutils. I'd rather fix it the right 
way, though, as the module is intended for more-general use.

Message #14 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: Mihai Moldovan <ionic <at> ionic.de>
To: 26044 <at> debbugs.gnu.org
Subject: Re: coreutils-8.27 fails to build
Date: Fri, 10 Mar 2017 21:19:19 +0100

[Message part 1 (text/plain, inline)]

We (naturally) also hit that in MacPorts.

Downstream bug report: https://trac.macports.org/ticket/53764

I've fixed it for our package by manually inserting the renameat gnulib module.

I assume that your change to bootstrap.conf also (automatically) adds the
renameat gnulib module, so that's indeed to right way to go.



Mihai

[signature.asc (application/pgp-signature, attachment)]

Message #19 received at 26044-done <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Mihai Moldovan <ionic <at> ionic.de>, 26044-done <at> debbugs.gnu.org
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Fri, 10 Mar 2017 21:28:54 -0800

On 10/03/17 12:19, Mihai Moldovan wrote:
> We (naturally) also hit that in MacPorts.
> 
> Downstream bug report: https://trac.macports.org/ticket/53764
> 
> I've fixed it for our package by manually inserting the renameat gnulib module.
> 
> I assume that your change to bootstrap.conf also (automatically) adds the
> renameat gnulib module, so that's indeed to right way to go.

Right. I had provided a tarball generated with the bootstrap.conf change
for the original reporter at:
https://www.pixelbeat.org/cu/coreutils-8.27.4-3c9d7.tar.xz
which you might find useful for comparison.

J has reported the build now works with the above change in effect,
so I've pushed and I'm marking this as done.

cheers,
Pádraig

Message #22 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: J Rogowsky <geometriker <at> gmail.com>, 26044 <at> debbugs.gnu.org
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Fri, 10 Mar 2017 23:11:03 -0800

[Message part 1 (text/plain, inline)]

On 09/03/17 23:22, Pádraig Brady wrote:
>> BTW, I had only one test FAIL in "make check":  tests/misc/sort-debug- 
>> keys
>> Is this related to the above?
> I'm guessing that "failure" might be locale related.
> Could you attach the tests/test-suite.log file generated with:
>   make TESTS=tests/misc/sort-debug-keys.sh SUBDIRS=. VERBOSE=yes check

Best keep all responses on the list.
The log you sent privately, did indeed indicate a locale issue.
In particular sort was operating in LC_ALL=C mode when
it should have been in fr_FR.utf8.  This is the only test
that sets LC_MESSAGES separately to LC_COLLATE etc.
so perhaps if we use LC_ALL for each sort invocation in
this test, it avoids this particular issue on your system?

Can you try running the above make check command again,
with the attached patch applied.

thanks,
Pádraig

[sort-debug-osx.patch (text/x-patch, attachment)]

Message #25 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: J Rogowsky <geometriker <at> gmail.com>
To: 26044 <at> debbugs.gnu.org
Cc: Pádraig Brady <P <at> draigBrady.com>
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Sat, 11 Mar 2017 10:12:09 -0700

[Message part 1 (text/plain, inline)]

On 2017.03.11, at 00:11:03, Pádraig Brady wrote:

> On 09/03/17 23:22, Pádraig Brady wrote:
>>> BTW, I had only one test FAIL in "make check":  tests/misc/sort- 
>>> debug-
>>> keys
>>> Is this related to the above?
>> I'm guessing that "failure" might be locale related.
>> Could you attach the tests/test-suite.log file generated with:
>>  make TESTS=tests/misc/sort-debug-keys.sh SUBDIRS=. VERBOSE=yes check
>
> Best keep all responses on the list.
Sorry. I am a newbie at this and not accustomed to doing this kind of  
thing.

> The log you sent privately, did indeed indicate a locale issue.
> In particular sort was operating in LC_ALL=C mode when
> it should have been in fr_FR.utf8.  This is the only test
> that sets LC_MESSAGES separately to LC_COLLATE etc.
> so perhaps if we use LC_ALL for each sort invocation in
> this test, it avoids this particular issue on your system?
>
> Can you try running the above make check command again,
> with the attached patch applied.
>
> thanks,
> Pádraig
> <sort-debug-osx.patch>

Yes, that solved the problem.
test results attached.

[testresults (application/octet-stream, attachment)]

[Message part 3 (text/plain, inline)]

Thank you for your efforts.
If there is anything I can do for you in the future, just let me know.

Regards,
J Rogowsky

Message #28 received at 26044 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: J Rogowsky <geometriker <at> gmail.com>, 26044 <at> debbugs.gnu.org
Subject: Re: bug#26044: coreutils-8.27 fails to build
Date: Sat, 11 Mar 2017 10:53:24 -0800

On 11/03/17 09:12, J Rogowsky wrote:
> On 2017.03.11, at 00:11:03, Pádraig Brady wrote:
> 
>> On 09/03/17 23:22, Pádraig Brady wrote:
>>>> BTW, I had only one test FAIL in "make check":  tests/misc/sort- 
>>>> debug-
>>>> keys
>>>> Is this related to the above?
>>> I'm guessing that "failure" might be locale related.
>>> Could you attach the tests/test-suite.log file generated with:
>>>  make TESTS=tests/misc/sort-debug-keys.sh SUBDIRS=. VERBOSE=yes check
>>
>> Best keep all responses on the list.
> Sorry. I am a newbie at this and not accustomed to doing this kind of  
> thing.
> 
>> The log you sent privately, did indeed indicate a locale issue.
>> In particular sort was operating in LC_ALL=C mode when
>> it should have been in fr_FR.utf8.  This is the only test
>> that sets LC_MESSAGES separately to LC_COLLATE etc.
>> so perhaps if we use LC_ALL for each sort invocation in
>> this test, it avoids this particular issue on your system?
>>
>> Can you try running the above make check command again,
>> with the attached patch applied.
>>
>> thanks,
>> Pádraig
>> <sort-debug-osx.patch>
> 
> Yes, that solved the problem.
> test results attached.

Excellent. Pushed at:
http://git.sv.gnu.org/gitweb/?p=coreutils.git;a=commitdiff;h=v8.27-6-g04148c9

thanks!
Pádraig.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.