On Sunday 07 November 2010, Ralf Wildenhues wrote: > Hello Stefano, Andy, > > * Stefano Lattarini wrote on Sat, Nov 06, 2010 at 06:10:44PM CET: > > On Friday 05 November 2010, Stefano Lattarini wrote: > > > I can confirm the bug with latest automake (from git master), with > > > a much-reduced minimal testcase (see attachment). > > > > > > I still haven't looked for an explanation or a fix, though. > > I've manged to find a very simple fix for the bug (see attached patch). > > > > OK to apply to maint? > > OK if it passes the whole testsuite. It does for me. BTW, I've also squashed a minor typofix in the new test's comments, and a better ChangeLog entry:
Fix a bug in variable concatanation with `+='. * lib/Automake/VarDef.pm (append): Since the content of the "appended-to" variable is going to be unconditionally normalized later, simply separate the appended value with a single whitespace character, instead of trying to be uselesssly smarter by using escaped newlines. This fixes a bug in which extra backslashes where erroneously inserted in the variable's final value. * tests/pluseq11.test: New test, exposing the bug. * tests/Makefile.am (TESTS): Update. Reported by Andy Wingo. Now, should I also try to close the bug #7333 on debbugs? The instruction at <http://debbugs.gnu.org/Developer.html> seems quite clear about how to do so, so I think I can manage to get it right. BTW, Ouch! I see that my previous reply presenting the patch has erroneously opended a new, spurious bug report (#7345) in the tracker! Ralf, could you please you close that report as invalid? Regards, Stefano