Re: 'make check' failure on MinGW

2009-05-24 Thread Bob Friesenhahn
On Sun, 24 May 2009, Ralf Wildenhues wrote: Yes, it is attached. Thank goodness it is so compressable. Thanks. Can you post the output of the following, done in an empty directory: I don't know if you still need this, but this is the output: echo .foo . .foo . echo .foo.log .log. .foo.log

Re: 'make check' failure on MinGW

2009-05-24 Thread Ralf Wildenhues
I've been able to reproduce this issue on GNU/Linux with GNU make 3.80. Committing this fix (will push out soon). An easy workaround for this issue is to ensure that the TESTS variable never contains trailing white space, i.e., the last test should be used unconditionally on every system. Thanks

Re: 'make check' failure on MinGW

2009-05-23 Thread Ralf Wildenhues
* Bob Friesenhahn wrote on Sat, May 23, 2009 at 11:52:22PM CEST: > On Sat, 23 May 2009, Ralf Wildenhues wrote: >> Can you post the output of >> make check SHELL="/bin/sh -x" >> >> on the failing system? > > Yes, it is attached. Thank goodness it is so compressable. Thanks. Can you post the outp

Re: 'make check' failure on MinGW (was: The New parallel-tests Framework)

2009-05-23 Thread Bob Friesenhahn
It seems that this problem is perhaps not make version specific since I installed GNU make 3.79.1 on a Solaris system and the test suite passes without the error. I did a diff between the MinGW Makefile and one from Solaris, and found not much interesting except for this: < am__test_logs2 =

'make check' failure on MinGW (was: The New parallel-tests Framework)

2009-05-23 Thread Ralf Wildenhues
* Bob Friesenhahn wrote on Sat, May 23, 2009 at 10:29:18PM CEST: > On Sat, 23 May 2009, Ralf Wildenhues wrote: >> * Bob Friesenhahn wrote on Sat, May 23, 2009 at 08:12:33PM CEST: >>> I now have parallel tests working in GraphicsMagick (verified under >>> Solaris, FreeBSD, and OS-X Leopard) except t