This commit applies a patch submitted by Nick Clifton to add these notices.
The email carrying the patch quoted a reply from the FSF copyright clerk
affirming that copyright for this file had been properly assigned to the FSF,
but the quoted text did not include a ticket number.
Tcl versions old enough to be incompatible with DejaGnu are very unlikely
to be encountered "in the wild" and there were ways for configure to find
one version of Tcl, but DejaGnu to actually run with a different instance
installed somewhere else on the system.
Further, testing for specific versions is generally discouraged in favor
of testing for the features actually used. Simply running the DejaGnu
testsuite will provide adequate assurance that the Tcl installation
found by the runtest script is sufficient for running DejaGnu.
This fixes "make distcheck" on my Debian 10 system.
Test I did:
autoreconf -vfi
./configure
make
make distcheck
ChangeLog:
* Makefile.am: Fix dejagnu.texi dependency.
Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
This reverts commit 1a76740eff.
This caused problems (reported in PR44462 as a secondary issue)
when a source tree is copied out of a Git directory. A better
solution will be developed for the 1.6.4 release that also tags
development versions when run from Git working directories without
relying on an installation hook.
The problem can be demonstrated with:
git -z ls-files | cpio -0dp /tmp/dejagnu-test-src
followed by configuring in that non-Git copy of the source tree.
Unrecognized `gnatmake' switches are not implicitly passed on to the
linker, so just pasting board `ldflags' and any other linker flags
verbatim into `add_flags' to use for the invocation line of `gnatmake'
will make them ignored at best.
For example in a GCC test environment that has:
set_board_info ldflags "-Wl,-dynamic-linker,.../sysroot/lib/ld-linux-riscv64-lp64d.so.1 -Wl,-rpath,.../sysroot/lib64/lp64d -Wl,-rpath,.../sysroot/usr/lib64/lp64d"
so that sysroot paths are correctly embedded with the binaries linked
for use with the dynamic loader and shared library dependencies, the
setting will be ignored for the GNAT test suite making all the execution
tests fail, e.g.:
PASS: gnat.dg/abstract_with_anonymous_result.adb (test for excess errors)
spawn qemu-riscv64 ./abstract_with_anonymous_result.exe
/lib/ld-linux-riscv64-lp64d.so.1: No such file or directory
FAIL: gnat.dg/abstract_with_anonymous_result.adb execution test
For `gnatmake' to pass switches on to the linker the `-largs' switch has
to be used, which affects all the switches that follow until a switch is
seen that changes the selection, like `-margs', which resets to the
initial state of the switch interpretation machine.
Wrap linker flags into `-largs'/`-margs' for Ada then, carefully
preserving the place these flags are placed at within `add_flags', as
surely someone will have depended on that, correcting test failures like
above:
PASS: gnat.dg/abstract_with_anonymous_result.adb (test for excess errors)
spawn qemu-riscv64 ./abstract_with_anonymous_result.exe
PASS: gnat.dg/abstract_with_anonymous_result.adb execution test
Pass multilib flags both to the compiler and to the linker as both build
stages interpret them.
Update the testsuite accordingly.
* lib/target.exp (default_target_compile): Wrap linker flags into
`-largs'/`-margs' for Ada.
* testsuite/runtest.libs/target.test (compile_test): Update
accordingly.
Signed-off-by: Maciej W. Rozycki <macro@wdc.com>