Richard Lowe wrote:
> Bill Sommerfeld wrote:
>> On Wed, 2007-07-04 at 00:55 -0700, bugzilla-daemon at np.grommit.com wrote:
>>> If $BRINGOVER_WS is set, $CODEMGR_WS is not (ie, a full bringover +
>>> build),
>>> nightly will bail out if $BRINGOVER_WS is not on the local
>>> filesystem, as it
>>> attempts to find usr/src/Makefile there for informational purposes:
>>>
>>> Unable to find "Makefile" in
>>> ssh://<hostname>//builds/onnv-clone/usr/src or
>>> /builds/work/onnv-sccs/this-wont-work/usr/src.
>>>
>>> This happens as it's trying to ouput tools versions, very early on,
>>> which it
>>> does before it brings over.
>>
>> there's a bunch of subtlety baked in there; see in particular:
>>
>> 6266590 nightly misbehaving when kicked-off from empty ws
>> 6232357 nightly can run through the air for hours before looking down
>>
>
> This latter one may have been re-broken by shadow compilation, or
> something close to it.
>
> cw seems to happily carry on if it only finds the shadow and not the
> primary, then error continually during the build.
>
> I'll file a bug.
>
CR 6577894 nightly spends hours chasing shadows if primary compiler path is
wrong
-- Rich
