t.valgrind problem

[David Relson]

On Sat, 20 Nov 2004 22:13:01 -0500
Charles A Edwards wrote:

> On Sat, 20 Nov 2004 22:02:17 -0500
> David Relson wrote:
> 
> > In the past, I've seen other oddities involving valgrind.  Running
> > as"relson" I've had t.valgrind fail, but running as "root" it
> > succeeds. I've not dug deeply to find out why, due to feeling out of
> > my depth and feeling that having the valgrind test work isn't of key
> > importance.
> 
> 
> See my subsequent post.
> 
> It may not be of 'key importance' but it does cause the build to fail.
> And as you also may be aware all Mdk rpm build are done as user,never
> as root.

Hi Charles,

No, I hadn't been aware that MDK rpm builds are done as user.  My
initial months of rpm builds were done as root 'cause I didn't know they
could be done as a user.  Since I figured out how to do the user builds,
that's all I've done.

As a developer, "make check" is run fairly frequently. I _hate_ it when
I release a patch that causes a regression.  For a while t.valgrind was
failing its test, but we're past that now.  IIRC, the problem was new
valgrind version and old options in our test.  While t.valgrind was
failing (as user), I accidentally discovered that it would succeed when
run as root.  I never did figure out why.

Regards,

David