Charles Wilson wrote:
> * libltdl/m4/libtool.m4 (_LT_SYS_DYNAMIC_LINKER): Fix handling
> of dos-style paths when parsing $CC -print-search-dirs output.
> ---
> Currently running regression tests on MSYS-1.0.12 + mingw-4.4.0-dw2 (TDM
> version).
>
> OK to push, if no regressions from earlier behavior on mingw?
On MSYS-1.0.12 with gcc-4.4.0dw2 (TDM):
=================================================================
OLD:
2 of 114 tests failed
(10 tests were not run)
FAIL: tests/f77demo-exec.test (after f77demo-static.test)
FAIL: tests/fcdemo-exec.test (after fcdemo-static.test)
NOTE: in this configuration, all of the other f77 and fc tests were
skipped. The real "bug" here is why these weren't skipped also; but I
don't think it has anything to do with this patch.
NEW:
82 tests behaved as expected.
6 tests were skipped.
However, a lot of these were reported as "expected failure" because my
MSYS-1.0.12 installation does not have any of the other autotools
installed. So, I also ran the test suite under the older setup:
On MSYS-1.0.11 with gcc-3.4.5 (mingw.org):
=================================================================
OLD:
4 of 106 tests failed
(18 tests were not run)
FAIL: tests/demo-exec.test [*]
FAIL: tests/f77demo-static.test [**]
FAIL: tests/f77demo-conf.test [**]
FAIL: tests/f77demo-shared.test [**]
[*] after demo-shared. This is fixed by one of my other recent patches.
[**] these failures have been around a while on mingw. Note the missing
g77.exe below:
checking for Fortran 77 libraries of g77... -L/usr/lib
-L/bin/../lib/gcc-lib/i686-pc-msys/2.95.3-1 -L/bin/../lib/gcc-lib -L/usr
/lib/gcc-lib/i686-pc-msys/2.95.3-1
-L/bin/../lib/gcc-lib/i686-pc-msys/2.95.3-1/../../../../i686-pc-msys/lib
-L/usr/lib/gcc-lib/i
686-pc-msys/2.95.3-1/../../../../i686-pc-msys/lib -lg2c -lmsys-1.0
-luser32 -lkernel32 -ladvapi32 -lshell32
checking for dummy main to link with Fortran 77 libraries... unknown
So, that's something else to fix, but it's not related to this patch AFAICT.
NEW:
8 failed (3 expected failures).
4 tests were skipped.
34: search-path.at:25 sys_lib_search_path
libtool
71: recursive.at:60 compiling softlinked libltdl
libtoolize autoconf automake
72: recursive.at:80 compiling copied libltdl
libtoolize autoconf automake
73: recursive.at:100 installable libltdl
libtoolize autoconf automake
86: cmdline_wrap.at:28 Run tests with low max_cmd_len
recursive
Now, 34 is ironic. However, it is because I don't have an accessible
zlib installed, so naturally the link never succeeds. But, it's at
least testing all the right directories:
34. search-path.at:25: testing ...
libtool: link: /mingw/src/libtool/libtool-2.2.7a/libltdl/config/compile
gcc -g -O2 -o .libs/main.exe main.o -Lc:/MinGW/lib/gcc/
mingw32/3.4.5 -lz
c:\MinGW\bin\..\lib\gcc\mingw32\3.4.5\..\..\..\..\mingw32\bin\ld.exe:
cannot find -lz
collect2: ld returned 1 exit status
Repeated with different -L options:
-Lc:/MinGW/lib/gcc/
-LC:/MinGW/lib/gcc/
-Lc:/MinGW/mingw32/
-LC:/MinGW/mingw32/
-L/mingw/lib
-Lc:/MinGW/lib -
-LC:/MinGW/lib
-L/lib
(actually, that directory list is *wrong* -- because mingw gcc 3.4.5's
specs file is wrong. It should never search /lib, that's msys stuff.
But...it does:
libraries:
=c:/MinGW/bin/../lib/gcc/mingw32/3.4.5/;c:/MinGW/bin/../lib/gcc/;C:/MinGW/lib/gcc/mingw32/3.4.5/;/usr/lib/gcc/mingw32
/3.4.5/;c:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../../../../mingw32/lib/mingw32/3.4.5/;c:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../..
/../../mingw32/lib/;C:/MinGW/lib/gcc/mingw32/3.4.5/../../../../mingw32/lib/mingw32/3.4.5/;C:/MinGW/lib/gcc/mingw32/3.4.5/../../.
./../mingw32/lib/;/mingw/lib/mingw32/3.4.5/;/mingw/lib/;c:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../../../mingw32/3.4.5/;c:/MinGW/b
in/../lib/gcc/mingw32/3.4.5/../../../;C:/MinGW/lib/gcc/mingw32/3.4.5/../../../mingw32/3.4.5/;C:/MinGW/lib/gcc/mingw32/3.4.5/../.
./../;/lib/mingw32/3.4.5/;/lib/;/usr/lib/mingw32/3.4.5/;/usr/lib/
So, the "bug" there is actually in gcc-3.4.5...
71:
C:\msys\1.0\local\bin\autoheader: line 159: exec:
C:\msys\1.0\local\bin\autoheader-2.61: not found
This is because 'the autoconf wrapper couldn't find the actual autoconf
with the specified version'. It's actually a wrapper script bug.
72:
same wrapper script bug.
73:
ditto.
86: just a repeat of #34 above.
=================================================================
OK to push?
--
Chuck
