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]

libstdc++ testsuite failures and GLIBCPP_BUILD_PCH


Testing of libstdc++-v3 for a cross powerpc64-linux compiler fails
because bits/stdc++.h isn't found.  It appears to be a configuration
or build problem, but I don't understand how that mechanism works.

>From libstdc++-v3/include/Makefile.am:

if GLIBCPP_BUILD_PCH
pch_build = ${pch_input}
pch_install = install-pch
else
pch_build =
pch_install =
endif

>From libstdc++-v3/include/Makefile.in:

@GLIBCPP_BUILD_PCH_TRUE at pch_build = ${pch_input}
@GLIBCPP_BUILD_PCH_FALSE at pch_build =
@GLIBCPP_BUILD_PCH_TRUE at pch_install = install-pch
@GLIBCPP_BUILD_PCH_FALSE at pch_install =

For a native i686-linux build, from
i686-pc-linux-gnu/libstdc++-v3/include/Makefile:

pch_build = ${pch_input}
#pch_build =
pch_install = install-pch
#pch_install =

For a cross build of powerpc64-linux from powerpc-linux, from
powerpc64-linux/libstdc++-v3/include/Makefile:

pch_build = ${pch_input}
pch_build =
pch_install = install-pch
pch_install =

So, what magic is going on with @GLIBCPP_BUILD_PCH_TRUE@ and
@GLIBCPP_BUILD_PCH_FALSE@, and why are they both true for my cross
build?

Janis


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