[rrd-developers] rrdtool-1.3.99909060808 configure error

Barrie barrie0482 at gmail.com
Sun Aug 23 08:24:36 CEST 2009


Hi Tobi,


The configure script is using the Solaris ctime_r test which has three 
parameters. The configure script needs to #include "src/win32comp.c" to 
access win32 ctime_r. The win32 ctime_r only has 2 parameters.
I'm sure it is all configurable in the configure.ac and the 
m4/acinclude.m4 scripts, but my skills with m4 and autoconf are worse 
than my C skills. :-)   Happy to learn. But I'm not sure where to start.

I have modified the configure script to work around this problem and the 
script continued until it failed with problems with dependencies. 
Working on that now.

A question. The configure script has an option to cache.  Is this 
possible ?  The  configure options  show  a  [DISABLED] against the 
--cache-file=FILE option. Does this mean the option is not available for 
the rrdtool configure script.

I have tried the -C and --config-cache options.

When the configure script runs I get the following.

configure: loading cache config.cache

The config.cache file is created, but it is zero length. It would be 
good to use as it takes so long to run the configure script through all 
the options that have passed.

Cheers, Barrie


Tobias Oetiker wrote:
> Friday Barrie wrote:
>
>   
>> Hi Tobi,
>>
>> I'm working with rrdtool-1.3.99909082100 now. I added the change noted  below
>> and that works OK. The configure script now makes it through the  IEEE  math
>> test.
>>
>> The next failure is at the ctime_r test. ctime_r for win32 is included in
>> win32comp.c in the src directory. I'm working on how configure can access that
>> function.
>>     
>
> Barry,
>
> how does configure fail on that ? it should only find that the
> function is missing and finally note the fact in the rrd_config.h
> file, which is fine since you can then use this finding to activate
> the appropriate replacement ...
>
> cheers
> tobi
>   
>> Cheers, Barrie
>>
>> Tobias Oetiker wrote:
>>     
>>> Hi Barrie,
>>>
>>> Today Barrie wrote:
>>>
>>>
>>>       
>>>> Hi BenBE,
>>>>
>>>> Thanks for the help. I didn't see your post come in, sorry for the late
>>>> reply.
>>>>
>>>>
>>>>         
>>>>> It uses the Solaris 10 branch, even though it should use none of them???
>>>>>
>>>>>
>>>>>           
>>>> I see the problem now. Just need to work out how to fix it. This is
>>>> proving to
>>>> be a good learning exercise. :-)
>>>>
>>>> /* solaris 10 it defines isnan such that only forte can compile it ... bad
>>>> bad
>>>> */
>>>> #if (defined(HAVE_ISNAN) && defined(isnan) && defined(HAVE_FPCLASS))
>>>> #  undef isnan
>>>> #  define isnan(a) (fpclass(a) == FP_SNAN || fpclass(a) == FP_QNAN)
>>>> #endif
>>>>
>>>> Thanks.
>>>>
>>>>         
>>> note that the way configure works is by trying many things out and
>>> when the fail going on to the next one ... it will take the one
>>> which works ...
>>>
>>> have a look at configure.ac and m4/acinclude.m4 how this is done
>>> ...
>>>
>>> in any event, to code above could be fixed like this:
>>>
>>> #if (defined(HAVE_ISNAN) && defined(isnan) && defined(HAVE_FPCLASS) &&
>>> defined(FP_SNAN) && defined(FP_QNAN))
>>> #  undef isnan
>>> #  define isnan(a) (fpclass(a) == FP_SNAN || fpclass(a) == FP_QNAN)
>>> #endif
>>>
>>> I have updated the trunk ... tomorrows snapshot will contain the
>>> change ...
>>>
>>> cheers
>>> tobi
>>>
>>>
>>>
>>>       
>>>> Cheers, Barrie
>>>>
>>>>
>>>>         
>>>>> Am 09.08.2009 11:52, schrieb Barrie:
>>>>>
>>>>>           
>>>>>> / checking if IEEE math works with signal
>>>>>>
>>>>>>             
>>>>> /The excerpt from the config.log is the following (if I found the
>>>>> correct
>>>>> place).
>>>>>
>>>>> configure:24927: checking if IEEE math works with signal(SIGFPE,SIG_IGN)
>>>>> configure:25019: gcc -o conftest.exe  -D_GNU_SOURCE -fno-strict-aliasing
>>>>> -Wall -std=c99 -pedantic -Wundef -Wshadow -Wpointer-arith -Wcast-align
>>>>> -Wmissing-prototypes -Wmissing-declarations -Wnested-externs -Winline
>>>>> -Wold-style-definition -W  -fPIC -DPIC
>>>>> -L/d/projects/rrdtool_mingw32/contrib
>>>>> -L/d/projects/rrdtool_mingw32/contrib conftest.c  >&5
>>>>> conftest.c:1: warning: -fPIC ignored for target (all code is position
>>>>> independent)
>>>>> conftest.c:116:5: warning: "HAVE_IEEEFP_H" is not defined
>>>>> conftest.c:120:5: warning: "HAVE_FP_CLASS_H" is not defined
>>>>> conftest.c: In function `main':
>>>>> conftest.c:169: warning: implicit declaration of function `fpclass'
>>>>> conftest.c:169: warning: nested extern declaration of `fpclass'
>>>>> conftest.c:169: error: `FP_SNAN' undeclared (first use in this function)
>>>>> conftest.c:169: error: (Each undeclared identifier is reported only once
>>>>> conftest.c:169: error: for each function it appears in.)
>>>>> conftest.c:169: error: `FP_QNAN' undeclared (first use in this function)
>>>>> configure:25023: $? = 1
>>>>> configure: program exited with status 1
>>>>>
>>>>> It uses the Solaris 10 branch, even though it should use none of them???
>>>>>
>>>>> Regards,
>>>>> BenBE.
>>>>>
>>>>>           
>>>   ------------------------------------------------------------------------
>>>
>>>
>>> No virus found in this incoming message.
>>> Checked by AVG - www.avg.com Version: 8.5.392 / Virus Database:
>>> 270.13.63/2316 - Release Date: 08/20/09 18:06:00
>>>
>>>
>>>       
>>     
>
>   
> ------------------------------------------------------------------------
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com 
> Version: 8.5.392 / Virus Database: 270.13.63/2316 - Release Date: 08/20/09 18:06:00
>
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/rrd-developers/attachments/20090823/b7b9bfb2/attachment.htm 


More information about the rrd-developers mailing list