Re: automake 1.5 fails 4 tests on Solaris

[Lars Hecking]

Tom Tromey writes:
> >>>>> "Lars" == Lars Hecking <address@hidden> writes:
> 
> Lars> checking whether  and cc understand -c and -o together... no
> Lars> checking for gcc... /tmp/automake-1.5/tests/testSubDir/compile 
> 
> This is interesting.  It suggests that `CC' is empty.
> Is that right?

 Yep, see below. And here's why: I reran the test suite on a Solaris 7 box
 (the OS release doesn't matter!), and these tests passed.

 The difference is: the Sol7 box has /usr/ucb/cc linked to gcc, whereas the
 Sol8 box hasn't. This would suggest that the test suite always tries to use
 cc, even if it's known not to work and gcc is available.

 Indeed: testSubDir/config.log says

configure:835: checking whether  and cc understand -c and -o together
configure:850: cc -c conftest.c -o conftest.o 1>&5
/usr/ucb/cc:  language optional software package not installed
configure:901: checking for gcc
configure:1014: checking whether the C compiler 
(/tmp/automake-1.5/tests/testSubDir/compile   ) works
configure:1030: /tmp/automake-1.5/tests/testSubDir/compile  -o conftest    
conftest.c  1>&5

 Is the problem here that AM_PROG_CC_C_O is called before AC_PROG_CC?