What if users warned about critical bugs?

[Thomas Ward]

I think we need to really tread carefully here...

On Mon, Nov 10, 2014 at 11:31 AM, Nio Wiklund <nio.wiklund at gmail.com> wrote:
> Den 2014-11-10 17:11, Alberto Salvia Novella skrev:
>> Thomas Ward:
>>> How would a user know what a critical bug is?
>>
>> Instead of asking to report critical bugs, you can ask "if the bug
>> causes data corruption or renders the system temporally or permanently
>> unusable, please warn about it to the Bug Control team".
>
> I think this is a good idea, definitely worth trying :-)

We need to be really careful with how we define 'data corruption'.
There are cases, such as in the nginx package, where data is
overwritten by the package because it ships defaults.  If the
configuration files, and/or included default file directories, get
overwritten, this can cause 'data corruption via overwriting', but
that's not a Critical bug, that's a case where the user used the
default location controlled by the package manager, not necessarily a
bug in the package itself.

If we define 'data corruption' to be, say, a partition-wide
corruption, that's different than a few configs being deleted or
corrupted but. being repairable or replaceable.  Therefore, we need to
explicitly define 'data corruption' in context of how we want
something to be determined as 'critical'.