Re: Results for egcs-2.92.11 19980921 (gcc2 ss-980609 experimental)

[Manfred Hollstein <manfred at s-direktnet dot de>]

On Mon, 28 September 1998, 03:41:35, law@cygnus.com wrote:

 > 
 >   In message <m0zNRH0-000395C@ocean.lucon.org>you write:
 >   > nowhere. The result is everyone wastes time on those bugs which can be
 >   > caught by libg++. Should that be a check-in requirement?
 > No.  It should not be a check-in requirement.
 > 
 > If we ever have check-in requirements, they would include either a
 > bootstrap of the compiler or running the existing testsuite, not
 > building/testing external code like libg++.
 > 
 > jeff

But, I agree with H.J., it would be fine if more people were testing
libg++ from time to time. For instance, the current mainline sources
look pretty much OK without libg++, but with libg++ "make check" fails
for hppa1.1-hp-hpux10.20, m68k-motorola-sysv and m88k-motorola-sysv3.
Perhaps it would also be helpful if we'd extend the test_summary script
to report if "configure"d directories have been found in the tree, but
the particular *.log files are missing. Currently if libg++ fails, no
results are posted for libg++ and nobody knows there's a problem :-(

manfred