This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Potential gcc-3.1 install problem


Hi, I was compiling gcc-3.1 on an old system, where `Makeinfo is too
old. Info documentation will not be built' according to configure.  I
was using gcc-2.95.3 to bootstrap, and all built fine.  So I did a
make install, which stopped with an error complaining that
gccint.info* is not found.  No problem, I edited the Makefile and run
make install again.  That's where all the trouble starts.  Somehow,
the changed Makefile triggered the recompilation of some files,
including toplev.c, however during make install, the default compiler,
i.e. 2.95.3 was used for that instead of the newly built compiler.  In
2.95.3 stdbool.h defines bool as an enum, so it has a size of 4, while
in gcc-3.1 bool is _Bool with size 1.  This changed the size of the
output_state struct, which caused toplev.c see the warning counts as
error counts, and halted all compiles that had any warnings (and it
would probably not halt compiles with errors.

I'm not sure if you would consider this a bug or user error.  It was
unexpected for me, and I would not be surprised if others would run
into this problem.  I did not open a formal bug on this, and I did not
spend time to figure out how to fix it.  I got my install working, and
I'm too busy to do anything more, but I just wanted to call your
attention to this issue.

Thanks,

Zoli


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]