Mathemagix

[Julien Wintz <address@concealed>]

It seems that it is a well known bug in some distributions of  
autoconf: see http://www.nabble.com/Weird-message-from-configure-t4748385.html 
 if interested.

The trick is to type rm -rf conftest* manually each time the  
compilation fails because of that (which should be done by autoconf  
but the r of rm -rf is missing in some verisons of it).

Here is the current compilation error:

 g++ -DHAVE_CONFIG_H -I../include -I../include -I../basix/include -I/ 
Users/jwintz/devel/mmx/include -I/usr/local/include -I/usr/local/ 
include/mathemagix -I/Users/jwintz/devel/mmx/include -ansi -Wall -O3 - 
fexpensive-optimizations -ftemplate-depth-35 -msse2 -MT libmmxlight_la- 
input.lo -MD -MP -MF .deps/libmmxlight_la-input.Tpo -c ../src/ 
input.cpp  -fno-common -DPIC -o .libs/libmmxlight_la-input.o
../src/input.cpp: In function 'char** mmx::complete(const char*, int,  
int)':
../src/input.cpp:279: error: 'rl_completion_matches' was not declared  
in this scope
../src/input.cpp: In function 'int mmx::passive_newline(int, int)':
../src/input.cpp:296: error: 'rl_crlf' was not declared in this scope
../src/input.cpp:297: error: 'rl_on_new_line' was not declared in this  
scope
../src/input.cpp:299: error: 'rl_done' was not declared in this scope
../src/input.cpp: In function 'int mmx::active_newline(int, int)':
../src/input.cpp:307: error: 'rl_crlf' was not declared in this scope
../src/input.cpp:308: error: 'rl_on_new_line' was not declared in this  
scope
../src/input.cpp:310: error: 'rl_done' was not declared in this scope
../src/input.cpp: In function 'bool  
mmx::shell_terminal_input(mmx::string&)':
../src/input.cpp:346: error: 'RETURN' was not declared in this scope
../src/input.cpp:347: error: 'META' was not declared in this scope

Ju.

On Nov 14, 2007, at 7:10 PM, Julien Wintz wrote:

> On Nov 14, 2007, at 6:26 PM, Grégoire Lecerf wrote:
>
> > > PS: after checking out I just ran ./configure, this may be not  
> > > sufficient ?
> > Yes it is.
> > Could you try the command "autoreconf" in basix/. and then try  
> > ".configure" again.
> > Grégoire.
>
> With an autoreconf in the root of mmx, I managed to build basix but  
> failed on mmxlight.
>
> With an autoreconf in the corresponding folders, I can see no  
> improvement.
>
> By the way, I still have strange messages like:
>
> checking if g++ PIC flag -fno-common works... rm: conftest.dSYM: is  
> a directory
> yes
> checking if g++ static flag -static works... rm: conftest.dSYM: is a  
> directory
> no
> checking if g++ supports -c -o file.o... rm: conftest.dSYM: is a  
> directory
> yes
> .... <- This part is ok
> checking for gettimeofday... yes
> disabling debugging
> enabling optimizations,  -ansi -Wall -O3 -fexpensive-optimizations
> enabling exception handling
> disabling correctness checks for low-level operations
> enabling glue compilation for interpreter
> disabling compilation of additional test programs
> disabling compilation of additional benchmark programs
> configure: creating ./config.status
> creating include/basix/basix-config.hpp - prefix BASIX for include/ 
> basix/config.hpp defines
> rm: conftest.dSYM: is a directory
> make[1]: *** [config.status] Error 1
> make: *** [basix-all] Error 2
>
> What is this "rm: conftest.dSYM: is a directory" doing in the middle  
> my configure test !!??!!
>
> Maybe the switch from X.4 to X.5 has something to do with this ... I  
> experience the same problem either using fink's autotools or  
> leopard's developer tools autotools.
>
> Ju.