[Rd] compiling tests/Embedding

[Prof Brian Ripley]

On Fri, 26 May 2006, Antonio, Fabio Di Narzo wrote:

> Hi all.
> The same compiler problem appears on my gcc-4.0.2 shipped with ubuntu
> breezy. Well fixed with B.R. indication.

Note: I already said

> > (There is a known problem here with gcc >= 4.0.0, with this fix in the
> > pre-2.3.1 sources.  We know that certain RedHat systems have gcc3's that
> > support visibility, not always completely.)

and it is believed to be fixed in the 2.3.1 RC versions now under test.

So we don't need any reports on gcc4 for 2.3.0: please try the current 
test version.  (That this appeared on gcc 3.3.4 was a surprise, given it 
does not on 3.3.5.)

>
> Antonio.
>
> 2006/5/26, Duncan Temple Lang <duncan at wald.ucdavis.edu>:
>>
>>
>> In addition to the compiler issues, the test should also be changed.
>> It should use R_tryEval() directly rather than Test_tryEval().  The test
>> preceeds the existence
>> of that exported routine that was motivated by the same
>> usage that gave rise to this test.
>>
>> I'll commit an update when I get a few minutes.
>>
>>   D.
>>
>> Prof Brian Ripley wrote:
>>> What system is this?
>>>
>>> It should work under gcc-3.3.4, as the issue is visibility attributes
>> and
>>> they are not supposed to be supported on that compiler.  (I suspect from
>>> the message that this is not a version of gcc released by GNU, a common
>>> source of problems.)  I happen to have an i686 FC3 system with a gcc
>> 3.3.5
>>> compiled from the GNU sources (to help track down a bug in code
>> generated
>>> by that compiler), and that does work in tests/Embedding.
>>>
>>> In src/main/context.c you will find
>>>
>>> context.c:Rboolean attribute_hidden R_ToplevelExec(void (*fun)(void *),
>>> void *data)
>>>
>>> and you need to remove 'attribute_hidden' and recompile.
>>>
>>> (There is a known problem here with gcc >= 4.0.0, with this fix in the
>>> pre-2.3.1 sources.  We know that certain RedHat systems have gcc3's that
>>> support visibility, not always completely.)
>
>>
>>>
>>> On Thu, 25 May 2006, George Ostrouchov wrote:
>>>
>>>> I am compiling the Embedding examples in the tests directory and get an
>>>> undefined reference. I include the make output as well as grep'd output
>>>> of nm on libR.so and compiler and arch information. Do I have an
>>>> improperly built R shared library or is there a problem with the
>>>> Embedding tests or something else I am not seeing?
>>>>
>>>> Thanks for any help!
>>>> George
>>>>
>>>>
>>>> ost at hawk0:/data5/ost/R/R-2.3.0/tests/Embedding> make
>>>> ../../bin/R CMD LINK gcc -o Rtest Rtest.o embeddedRCall.o -L`cd ../..
>> &&
>>>> /bin/pwd`/lib -lR
>>>> gcc -o Rtest Rtest.o embeddedRCall.o
>>>> -L/autofs/tewa_data5/ost/R/R-2.3.0/lib -lR -Wl,--rpath
>>>> -Wl,/data5/ost/R/R-2.3.0/lib
>>>> embeddedRCall.o(.text+0x1e): In function `Test_tryEval':
>>>> /autofs/tewa_data5/ost/R/R-2.3.0/tests/Embedding/embeddedRCall.c:81:
>>>> undefined reference to `R_ToplevelExec'
>>>> collect2: ld returned 1 exit status
>>>> make: *** [Rtest] Error 1
>>>> ost at hawk0:/data5/ost/R/R-2.3.0/tests/Embedding> cd
>>>> /autofs/tewa_data5/ost/R/R-2.3.0/lib
>>>> ost at hawk0:/autofs/tewa_data5/ost/R/R-2.3.0/lib> nm libR.so | grep R_Top
>>>> 000000000042c920 B R_Toplevel
>>>> 000000000042cae8 B R_ToplevelContext
>>>> 000000000007a7b0 t R_ToplevelExec
>>>> ost at hawk0:/autofs/tewa_data5/ost/R/R-2.3.0/lib> gcc --version
>>>> gcc (GCC) 3.3.4 (pre 3.3.5 20040809)
>>>> Copyright (C) 2003 Free Software Foundation, Inc.
>>>> This is free software; see the source for copying conditions.  There is
>> NO
>>>> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR
>> PURPOSE.
>>>>
>>>> ost at hawk0:/autofs/tewa_data5/ost/R/R-2.3.0/lib> arch
>>>> x86_64
>>>> ost at hawk0:/autofs/tewa_data5/ost/R/R-2.3.0/lib>
>>>>
>>>>
>>>>
>>>
>>
>> ______________________________________________
>> R-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>
>
> Antonio, Fabio Di Narzo.
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595