Confirming/Closing GNATS reports

[Joseph S. Myers]

On Tue, 6 May 2003, Dara Hazeghi wrote:

> substitute for testing a native build. Also, is it
> considered reasonable just to test 3.3 on these sort
> of target specific bugs, or does one need to test
> against mainline too?

Where the bug is an --enable-checking failure (tree check, RTL check error
messages mainly) testing against mainline (or a --enable-checking build of
3.3 branch) is necessary - mainline has --enable-checking on by default,
release branches don't.  Testing mainline is a good idea anyway, in case
the problem has regressed, though for cross-compiler tests it's more
effort than for native builds for which you just keep several different
compiler version installations about.

-- 
Joseph S. Myers
jsm28@cam.ac.uk