This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: Disregard disabled languages
The script I was using passed:
--enable-languages=c,c++ --disable-bootstrap
to "contrib/gcc_build build". I believe that gcc_build does a "make
bootstrap". So, that's a weird operating mode.
gcc_build may do just a "make" now. Then it will do the right thing
independent of whether "--disable-bootstrap" is passed or not (I am not
as picky as Dan on the validity of this combination). Still, remember
that C++ patches need to be bootstrapped anyway, or from time to time
we'll break bootstrap because of an uninitialized variable that the host
compiler does not detect.
Yes, my patch removed that ability. (I hadn't imagined such a brave
feature.)
It's the toplevel bootstrap equivalent of "make f951 rebuilds files in
the Fortran front-end without optimization so that I can debug my
changes properly". The difference is that a make bubblestrap from the
toplevel will rebuild everything with the proper optimization flags and
bootstrapping compiler, to avoid the problems described above.
It's the typical feature whose presence you don't realize -- it's really
quite automatic to invoke it:
<hmm, there's a bug... let's debug it>
make all-stage1 (or make stage1-start, it is the same)
cd gcc
...
<ok, now it should work>
make
make: Nothing to be done for `all'.
<huh, right, stage1 does not have fortran enabled>
make f951
make: *** No rule to make target `f951'. Stop.
<what do I do now?>
I'm not sure what to do; it looks like I broke a feature, but removed a
bug. Should I revert my patch?
What I'd suggest is investigating whether gcc_build works if it does
"make" instead of "make bootstrap". Whether to revert the patch before
or after that, it's your call -- IANAM (I'm not a maintainer), and
neither is Dan, but we happen to agree that it's wrong. :-)
Paolo