[SCM] systemtap: system-wide probe/trace tool branch, master, updated. release-1.2-446-g64cddf3

[Josh Stone]

On 07/09/2010 03:43 AM, Mark Wielaard wrote:
> BTW. I am also seeing the following failure (both i686 and x86_64):
> FAIL: systemtap.printf/ptr.stp
> The new code seems to give a different result for "%-20pX\n".

Ugh.  That test is codifying some nasty behavior which I would rather do
away with.  Such as:

1. the incoming value is truncated from int64_t to unsigned long.
2. the precision is completely ignored
3. it uses an effective precision of 2*sizeof(void*), unless that would
   exceed the given positive field_width
4. negative field_width is treated exactly like a precision

That's WAY too much special behavior, IMO.  For now, I've only added the
default precision = 2*sizeof(void*) for --compatible=1.2, but I could
add the rest too.

This also raises to mind that all of our current testing is on 1.2's old
behavior, but when we reach the release we'll be at 1.3.  Frank
mentioned that we should probably start bumping the version soon after
each release, going forward.

Josh