R devel - Oct 2002 - configure/build issue with gcc 3.2.0 (PR#2176)

There is an incompatibility between R's configure/build process, and gcc
3.2.0 (and maybe earlier).
Gcc 3.2.0 includes -I/usr/local/include by default, on my system (default
configuration), and cpp0
issues a warning when this flag is issued additionally by the user (= R's
configure). This warning
ends up in the dependency sections of the Makefile that are generated
"automagically". The 1st
occurrence is line 105 of src/appl/Makefile .

The only way around it that I found, other than removing all
-I/usr/local/include in all Makefiles is to manually add
-Wp,-w, a gcc/cpp argument that suppresses all cpp warnings. It also has the
effect of not aborting
a compilation as soon as a specified header file is *not* found :(
If you decide to leave -I/usr/local/include and add -Wp,-w for gcc 3 compilers,
I would suggest
*not* adding -Wp,-w if -fsyntax-only is also given, to not interfere with
requested syntax-checking.
(that would probably require to call the compiler from a wrapper script,
something I already do.)

Rene Bertin

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

This has been discussed on R-devel, and it is definitely not an R bug
as there are workarounds stated in the R-admin manual.

Setting CPPFLAGS= (nothing) does avoid this except in the gnome module. It
was intended to put that explicitly in the R-admin manual for 1.6.0, but
it was not done in time (it is now there for R-patched).

We are told that this is fixed in the alphas of gcc 3.3.  Once it is clear
exactly which versions of gcc are affected we may add a workaround in the
configure script.

On Wed, 16 Oct 2002 rjvbertin@hotmail.com wrote:
> There is an incompatibility between R's configure/build process, and
gcc 3.2.0 (and maybe earlier).
> Gcc 3.2.0 includes -I/usr/local/include by default, on my system (default
configuration), and cpp0
> issues a warning when this flag is issued additionally by the user (=
R's configure). This warning
> ends up in the dependency sections of the Makefile that are generated
"automagically". The 1st
> occurrence is line 105 of src/appl/Makefile .
>
> The only way around it that I found, other than removing all
-I/usr/local/include in all Makefiles is to manually add
> -Wp,-w, a gcc/cpp argument that suppresses all cpp warnings. It also has
the effect of not aborting
> a compilation as soon as a specified header file is *not* found :(
> If you decide to leave -I/usr/local/include and add -Wp,-w for gcc 3
compilers, I would suggest
> *not* adding -Wp,-w if -fsyntax-only is also given, to not interfere with
requested syntax-checking.
> (that would probably require to call the compiler from a wrapper script,
something I already do.)
>
> Rene Bertin
>
>
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> r-devel mailing list -- Read
http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
>
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
>
-- 
Brian D. Ripley,                  ripley@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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Thu, 17 Oct 2002, RenE J.V. Bertin wrote:
> On Wed, 16 Oct 2002 21:01:30 +0100 (BST), <ripley@stats.ox.ac.uk>
wrote regarding "Re:
> configure/build issue with gcc 3.2.0 (PR#2176)"
>
> 8-) This has been discussed on R-devel, and it is definitely not an R bug
> 8-) as there are workarounds stated in the R-admin manual.
>
> 	Really? I have looked around the occurrences of 'gcc',
'cpp' and
'workaround' (none) and '-Wp' (none) in doc/manual/R-admin.pdf,
and didn't see anything that looked appropriate. Are we talking
about the same issue (not a bug *in* R indeed)?

Please do read *all* of my message.  I said that setting CPPFLAGS to empty
solves the problem.  Setting CPPFLAGS appropriately *is* discussed in that
manual, and that's what you neglected to do.

-- 
Brian D. Ripley,                  ripley@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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._