Re: Testsuite failures on UnixWare 7.1.1

[Akim Demaille]

| Alexandre Oliva wrote:
| > 
| > On Jan 26, 2001, Matt Schalit <address@hidden> wrote:
| > 
| > > I'm concerned that the CVS version will have more bugs in it.
| > > If it's not stable, then I thought it's not stable.
| > 
| > Maybe using yesterday's candidate release, just taken out of the CVS
| > tree, would make you feel better?  We'd rather hash out any bugs
| > before the release :-) That's why I suggested you might want to try
| > the CVS tree...
| > 
| > --
| > Alexandre Oliva
| 
| 
| Ok here's what happened when I tried the cvs checkout autoconf
| on Friday night:
| 
|   I moved my gcc to gcc.orig, so autoconf would compile with
|   /usr/bin/cc.  (The reason is I'm having a problem where any
|   configure script that tries to use gcc also tries to use
|         /usr/local/i586-sco-sysv5uw7.1.1/bin/as
|   instead of /usr/bin/as.  The /usr/local version of as doesn't
|   understand -Qy.
| 
|   Using ./configure, cc to compile, and gnu make to make.
| 
|   gmake check failed one test:
| 
| ========================================
| Testing suite log for GNU Autoconf 2.49d
| ========================================
| 
| Failed tests:
|   37: compile.at:205  GNU Fortran 77
| 
| Skipped tests:
|    9: tools.at:411    autoupdate
|   10: tools.at:447    autoupdating AC_LINK_FILES
|   11: tools.at:476    autoupdating AC_PREREQ
|   50: foreign.at:9    Autoconf & Libtool
| 
| 
| [snip]
| 
| configure:974: checking for pgf90
| configure:999: result: no
| configure:974: checking for fc
| configure:996: result: fc
| configure:1009: checking for object suffix
| configure:1020: fc -c  conftest.f >&5
| /usr/bin/fc[14]: fc: -c: unknown option
| Usage: fc [-lnrs] [-e editor] [-N #] [first] [last]
| configure:1023: $? = 2
| configure: failed program was:
|       program main
| 
|       end
| configure:1035: error: cannot compute OBJEXT: cannot compile

That was to be expected... I didn't like too much the changing of the
order between computing EXEEXT/OBJEXT and _WORKS, since it is the
later that exits ``gracefully'' when the compiler is broken (exit 77).

We have to do something about it, Tim.