Re: 01/01: gnu: custom-gcc: Delete broken or conflicting executables.

[Ricardo Wurmus]

Ricardo Wurmus <address@hidden> writes:

> Efraim Flashner <address@hidden> writes:
>
>> https://lists.gnu.org/archive/html/guix-commits/2016-03/msg00229.html
>>
>> This patch seems to have broken slepc[1], slepc-complex[2] and
>> randomjungle[3].
>>
>>
>> [1] http://hydra.gnu.org/build/1084459
>> [2] http://hydra.gnu.org/build/1084166
>> [3] http://hydra.gnu.org/build/1084474
>
> For randomjungle the config.log tells me this:
>
>> configure:19068: gcc -o conftest -g -O2  -I/usr/include  -L/usr/lib -lz 
>> conftest.c -lgsl -lm -lgslcblas -lxml2 -lpthread -lgfortran -lz  >&5
>> ld: cannot find -lgfortran
>> collect2: error: ld returned 1 exit status
>> configure:19068: $? = 1
>> configure: failed program was:
>> | /* confdefs.h */
>> | #define PACKAGE_NAME "randomjungle"
>> | #define PACKAGE_TARNAME "randomjungle"
>> | #define PACKAGE_VERSION "2.0.0"
>> | #define PACKAGE_STRING "randomjungle 2.0.0"
>> | #define PACKAGE_BUGREPORT ""
>> | #define PACKAGE_URL ""
>> | #define PACKAGE "randomjungle"
>> | #define VERSION "2.0.0"
>> | #define STDC_HEADERS 1
>> | #define HAVE_SYS_TYPES_H 1
>> | #define HAVE_SYS_STAT_H 1
>> | #define HAVE_STDLIB_H 1
>> | #define HAVE_STRING_H 1
>> | #define HAVE_MEMORY_H 1
>> | #define HAVE_STRINGS_H 1
>> | #define HAVE_INTTYPES_H 1
>> | #define HAVE_STDINT_H 1
>> | #define HAVE_UNISTD_H 1
>> | #define HAVE_DLFCN_H 1
>> | #define LT_OBJDIR ".libs/"
>> | #define HAVE_MATH_H 1
>> | #define HAVE_STDLIB_H 1
>> | #define HAVE_STRING_H 1
>> | #define HAVE_SYS_TIME_H 1
>> | #define HAVE_GETOPT_H 1
>> | #define HAVE__BOOL 1
>> | #define HAVE_STDBOOL_H 1
>> | #define SIZEOF_VOID_P 8
>> | #define HAVE_LIBZ 1
>> | #define HAVE_BOOST_DYNAMIC_BITSET_HPP 1
>> | /* end confdefs.h.  */
>> | #include <error.h>
>> | int
>> | main ()
>> | {
>> | error_at_line (0, 0, "", 0, "an error occurred");
>> |   ;
>> |   return 0;
>> | }
>> configure:19076: result: no
>
> I’ll investigate this.  It’s odd that “-lgfortran” cannot be found only
> because we deleted useless wrappers.

I found that for randomjungle the native inputs must now also include

    ("gfortran" ,gfortran "lib")

in *addition* to the usual

    ("gfortran" ,gfortran)

I can prepare a patch for the three broken packages, but I wonder why
this is necessary.  “gfortran” alone does not install “libgfortran.so”,
only the “lib” output does.

~~ Ricardo