igraph-help
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [igraph] ./configure won't detect GMP


From: Szabolcs Horvát
Subject: Re: [igraph] ./configure won't detect GMP
Date: Thu, 3 Sep 2015 13:35:52 +0200

Hi Tamas,

I just figured it out.  The problem was that I set CFLAGS but it seems
I need to set CPPFLAGS instead.  CFLAGS won't work.

The reason why it seemed to work with MacPorts gmp is that actually I
was using a MacPorts compiler at that time, which does look in
/opt/local/include by default.

On 3 September 2015 at 13:30, Tamas Nepusz <address@hidden> wrote:
> Hi Szabolcs,
>
> Can you send me the file named "configure.log" (to my private address)?
>
> T.
> T.
>
>
> On Thu, Sep 3, 2015 at 1:18 PM, Szabolcs Horvát <address@hidden> wrote:
>> Hi Tamas,
>>
>> Actually gmp.h is present in $HOME/local/include.  I'm really confused
>> about what is going wrong now.  Maybe there's something obvious I'm
>> missing?
>>
>> Witness:
>>
>> $ ls $HOME/local/include
>> gmp.h  igraph
>>
>> This is part of the output from ./configure:
>>
>> checking for libxml/parser.h... yes
>> checking for __gmpz_add in -lgmp... yes
>> checking gmp.h usability... no
>> checking gmp.h presence... no
>> checking for gmp.h... no
>> checking that generated files are newer than configure... done
>>
>> So you are right that it doesn't find gmp.h
>>
>> But if I crudely remove the checking part from configure.ac and force
>> it to try to use GMP anyway by setting gmp_support=yes, then igraph
>> compiles without errors and works fine.
>>
>> On 2 September 2015 at 22:31, Tamas Nepusz <address@hidden> wrote:
>>> Hi Szabolcs,
>>>
>>> The configure script looks for a header named gmp.h and a library that
>>> contains the symbol named __gmpz_add. I think that your LDFLAGS is
>>> okay since it contains a .dylib or .so file that contains __gmpz_add,
>>> but your CFLAGS variable should be like this:
>>>
>>> export CFLAGS=-I$HOME/local/include/gmp
>>>
>>> This is because gmp installs its headers in $prefix/include/gmp and
>>> not in $prefix/include, and igraph tries to include them with #include
>>> <gmp.h>.
>>>
>>> T.
>>>
>>> T.
>>>
>>>
>>> On Wed, Sep 2, 2015 at 5:54 PM, Szabolcs Horvát <address@hidden> wrote:
>>>> Dear All,
>>>>
>>>> I'm compiling igraph on OS X 10.10.5 using the system compiler.
>>>>
>>>> I'm trying to use GMP 6.0.0 from https://gmplib.org/, which I
>>>> installed into $HOME/local.  I configure igraph like this:
>>>>
>>>> export CFLAGS=-I$HOME/local/include
>>>> export LDFLAGS=-L$HOME/local/lib
>>>> ./configure --prefix=$HOME/local
>>>>
>>>> The configure script fails to detect GMP.  How can I get it to succeed?
>>>>
>>>> Notes: It does detect MacPorts's GMP if I try to use it.  It's also
>>>> version 6.0.0.  If I remove the part of the configure script that
>>>> detects GMP to trick it into thinking that it's present, igraph does
>>>> compile correctly, and it does work correctly (including functions
>>>> that use GMP).
>>>>
>>>> I don't know anything about autoconf and don't understand how it tries
>>>> to detect GMP ...
>>>>
>>>> Am I doing something wrong or is the GMP detection buggy?
>>>>
>>>> Szabolcs
>>>>
>>>> _______________________________________________
>>>> igraph-help mailing list
>>>> address@hidden
>>>> https://lists.nongnu.org/mailman/listinfo/igraph-help
>>>
>>> _______________________________________________
>>> igraph-help mailing list
>>> address@hidden
>>> https://lists.nongnu.org/mailman/listinfo/igraph-help
>>
>> _______________________________________________
>> igraph-help mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/igraph-help
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help



reply via email to

[Prev in Thread] Current Thread [Next in Thread]