[security-dev 00295]: Re: Heads up: Bad changeset rollback

[Brad Wetmore]

The current policy is still one putback per bugid.  There have been 
other really good reasons to allow more than one (pre-built JCE 
binaries), but so far no change in the policy.  I still do wish it were 
changed, as filing a separate bug just to do a JCE build is wasted work, 
IMHO.

 > Although I guess you
 > could reuse the swapped bug number for that one then :)

We think about hijacking that bug.  ;)  We also considered a new 
bug/changeset combo that said "1234567:  changeset ffffffffffff should 
list bug aaaaaaaaaaaa".  Since only a well defined and relatively small 
set of people are pulling the JSN gate, and those people were warned 10 
minutes after it happened, we thought a rollback was best.  If it 
happened in the MASTER, we probably would have gone the other way.

Brad


Mark Wielaard wrote:
> On Thu, 2008-09-04 at 20:23 +0800, Max (Weijun) Wang wrote:
>> The bug id should have been 6740833.
> 
> Aha! The 7 and 4 got swapped. Thanks. And the jcheck scripts would then
> have prevented that bug to ever get fixed again. Although I guess you
> could reuse the swapped bug number for that one then :)
> 
> It might be an idea to add some kind of override possibility to the
> commit message to say "yes, I know this bug number has been used before,
> but this commit is the real fix". Maybe just add a + to the bug number
> to reuse one?
> 
> Cheers,
> 
> Mark
>