`configure' Problem: Building within the GCC Tree
bernd.edlinger at hotmail.de
Thu Dec 4 22:55:05 UTC 2014
On Wed, 3 Dec 2014 10:39:57, Torbjörn Granlund wrote:
> I am not aware of any problems except for the ones reported on relation
> to GCC in the last few weeks. Our very broad internal testing has never
> struck any lex problem.
I mean this https://gmplib.org/list-archives/gmp-bugs/2010-December/002130.html:
it can be easily reproduced like this (assume a working flex is installed):
trap "" SIGPIPE
../gmp-4.3.2/configure --host=none --target=none
checking for tputs in -lncurses... no
checking for tputs in -lcurses... no
checking for readline in -lreadline... no
checking readline detected... no
checking for bison... bison -y
checking for flex... flex
checking lex output file root... configure: error: cannot find output from flex; giving up
> I don't know which implications splitting out a new configure.ac into
> the GMP demos directory will have. And I'm afraid I don't have time to
> work on this project.
I understand, and GCC will not depend on it, because I found a way how to work around
these probems even for gmp-4.3.2, that does not need any changes in the configure
script. In case you are curious, the trick is adding LEX="touch lex.yy.c" to the command
line. Well, I know it's a hack, but it works.
Regarding the splitting, I have half of the work already done. That is removing
all demo stuff from configure.ac and Makefile.am, that was quite simple.
But I had no sample how to write a GMP application autoconfig project from scratch.
Therefore I think the demo programs are nice to have, but a sample how to write an
autoconfig project that is linking with the GMP library would be a real added value.
> Please encrypt, key id 0xC8601622
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 3987 bytes
Desc: not available
More information about the gmp-discuss