Re: Backports bugs in the BTS
On 8 May 2017, at 12:56, Adrian Bunk <bunk@debian.org> wrote:
> On Mon, May 08, 2017 at 10:12:29AM +0000, Holger Levsen wrote:
>> On Sun, May 07, 2017 at 07:07:26PM +0200, Julian Andres Klode wrote:
>>> If at all in the existing bug report, but usually bug reports for
>>> backports are supposed to be handled on the mailing list.
>>
>> well, thats the default (which I disagree with / object to, but thats another
>> story) but it's totally ok to file bugs in the BTS for a package as
>> *as long as the packages maintainers are fine with that*.
>
> You don't need the permission of the maintainer to file a bug.
>
> And users do report such issues as bugs, e.g. this botched backport was
> discovered when searching for the root cause of #860147.
>
>> I do it for my packages all the time / whenever suitable. I like the BTS
>> and being able to properly track bugs. A mailinglist is a sink.
>
> Being able to track bugs is not the same as actually doing tracking.
>
> I ran into #860147 when going through the RC bugs that do according to
> the version tracking information not apply to unstable.
>
> Looking at all bugs, this is actually not that uncommon:
>
> udd=> SELECT COUNT(*) FROM bugs_found_in WHERE version LIKE '%~bpo%' AND
> NOT (id IN (SELECT id FROM bugs_found_in WHERE version NOT LIKE
> '%~bpo%')) AND NOT (id IN (SELECT id FROM bugs WHERE status = 'done'));
> count
> -------
> 453
>
> In other words, there are currently 453 open bugs in the BTS reported
> against backports and not marked as found in any non-backport version.
>
> For RC bugs it actually matters that the non-bpo version has to be added
> as found if the bug is not only in the backport - otherwise the bug
> doesn't show up in RC bugs lists.
>
> And after marking all bugs that are not specific to the backport as
> found in the non-bpo version, the query would only show broken backports
> like #860147 or #854610.
>
> Version tracking allows easy tracking of issues in the BTS,
> but for being effective this cannot be a per-maintainer choice.
>
> None of the above requires that the BTS knows anything about backports.
If it's a backports-specific bug, then yes, you should have the permission of
the maintainer (who may not even be the backporter of course) - the
instructions for reporting bugs specifically state "Please report bugs that you
found in the packages to the backports mailing list and NOT to the Debian
BTS!"[1], so that must be your default course of action.
(Just because I said the above does not mean I agree with the policy)
Regards,
James
[1] https://backports.debian.org/Instructions/#index6h2
Reply to: