bug-gnulib@gnu.org mirror (unofficial)
 help / color / mirror / Atom feed
From: Mohammad Akhlaghi <mohammad@akhlaghi.org>
To: Bruno Haible <bruno@clisp.org>, bug-gnulib@gnu.org
Cc: Elham Saremi <saremi@ipm.ir>
Subject: Re: getgroups.c failes to compile
Date: Fri, 17 May 2019 20:50:28 +0200	[thread overview]
Message-ID: <fcca5484-3d04-9ff3-7615-c5417a3be502@akhlaghi.org> (raw)
In-Reply-To: <2260855.Gdx5C7bdKa@omega>

Hi Bruno,

Thank you very much for the great explanation and solution. The 
elaborations really made things clear and were very useful. We applied 
the second one (setting `rpath'), and the problem was solved.

To avoid similar situations in the future, it may be good to add a check 
in the `configure' script so if `getgroups' fails, but the system is a 
GNU system, the configure script crashes with a warning (after all 
`make' will crash, but a crash in `configure' is more clear than a crash 
in `make'). Of course, this test can be done with all components similar 
to `getgroups' through a single macro/function.

Thanks again for the great and very useful solution,
Cheers,
Mohammad

On 5/14/19 3:35 AM, Bruno Haible wrote:
> Hi,
> 
> Mohammad Akhlaghi wrote:
>> Recently in a build of Gnuastro 0.9 (which uses Gnulib
>> `v0.1-2539-gd6af24178'), we confronted a crash during the build of
>> Gnulib's `getgroups.c' (error summary is in P.S.).
> 
> A log of the command "make" would be more useful than a log of the command
> "make -j8", because the latter intermingles the output of different
> compilations running in parallel. For example, here it looks like the
> error messages are coming from the compilation of getopt1.c:
> 
>> /bin/sh ../../libtool  --tag=CC   --mode=compile /usr/bin/gcc
>> -DHAVE_CONFIG_H -I. -I../..   -I/usr/local/include  -Wall -O3  -pthread
>> -MT getopt1.lo -MD -MP -MF $depbase.Tpo -c -o getopt1.lo getopt1.c &&\
>> mv -f $depbase.Tpo $depbase.Plo
>> In file included from getgroups.c:22:0:
>> getgroups.c:33:1: error: conflicting types for ‘rpl_getgroups’
>>    getgroups (int n _GL_UNUSED, GETGROUPS_T *groups _GL_UNUSED)
>>    ^
>> ./unistd.h:1216:1: note: previous declaration of ‘rpl_getgroups’ was here
>>    _GL_FUNCDECL_RPL (getgroups, int, (int n, gid_t *groups));
>>    ^
>> Makefile:2208: recipe for target 'getgroups.lo' failed
> 
>> As further information, I am also attaching several outputs of the
>> configuration and build which will hopefully provide much more detailed
>> information on the host system and build environment. The two
>> `redirect-*.txt' files are the full outputs of the `./configure'
> 
> These logs are more interesting:
> 
>> checking for working getgroups... no
> 
> On glibc systems, getgroups is expected to work. So, the cause is already
> to be found during the autoconfiguration.
> 
>> configure:23138: checking type of array argument to getgroups
>> configure:23172: /usr/bin/gcc -o conftest -Wall -O3  -pthread -I/usr/local/include -L/usr/local/lib conftest.c -ltiff -llzma -ljpeg -lwcs -lcfitsio -lz -lgsl -lgslcblas -lm   >&5
>> configure:23172: $? = 0
>> configure:23172: ./conftest
>> ./conftest: error while loading shared libraries: libwcs.so.6: cannot open shared object file: No such file or directory
>> configure:23172: $? = 127
>> configure: program exited with status 127
> 
> The error comes from the fact that you have a -L option, the directory in this
> option was actually used to locate a shared library (libwcs.so.6 in this case),
> but the runtime linker cannot find this shared library.
> 
> There are two easy workarounds:
>    (a) set the environment variable LD_LIBRARY_PATH=/usr/local/lib
>        before the configuration,
>    (b) pass LDFLAGS="-L/usr/local/lib -Wl,-rpath,/usr/local/lib".
> 
> In the case (a) the binaries that you build will only work when
> LD_LIBRARY_PATH=/usr/local/lib is still set. In the case (b) the
> binaries will work without this environment variable.
> 
> Bruno
> 


  reply	other threads:[~2019-05-17 18:48 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-05-11 22:06 getgroups.c failes to compile Mohammad Akhlaghi
2019-05-14  1:35 ` Bruno Haible
2019-05-17 18:50   ` Mohammad Akhlaghi [this message]
2019-05-18  0:58     ` Bruno Haible
2019-05-18 22:30       ` Mohammad Akhlaghi
2019-05-18 23:14         ` Bruno Haible
2019-05-19 21:15           ` Mohammad Akhlaghi
2019-05-23 16:17             ` Mohammad Akhlaghi
2019-05-25 23:52               ` Bruno Haible
2019-05-26 23:48                 ` Mohammad Akhlaghi
2019-05-27  7:17                   ` Bruno Haible

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://lists.gnu.org/mailman/listinfo/bug-gnulib

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=fcca5484-3d04-9ff3-7615-c5417a3be502@akhlaghi.org \
    --to=mohammad@akhlaghi.org \
    --cc=bruno@clisp.org \
    --cc=bug-gnulib@gnu.org \
    --cc=saremi@ipm.ir \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).