On 2016-10-08 15:17, Boris Zbarsky wrote:
On 10/8/16 11:39 AM, Filipus Klutiero wrote:
2. The ticket in which I reported that bugs.mozilla.org allows
tickets to be marked as both WONTFIX and resolved
In Bugzilla, a bug can be in multiple states that indicate what work remains to be done on it: UNCONFIRMED,
NEW, RESOLVED (and REOPENED, which I think we should get rid of). Those correspond to "need to check
whether this is a real issue", "need to deal with the issue", and "no more work needs to
be done here".
The RESOLVED state has different resolutions that explain _why_ no more work
needs to be done. These include FIXED (issue is fixed), INVALID (behavior is
as intended, e.g. because it's required by a specification), and WONTFIX
(acknowledgement that this is an issue, but an explicit decision that the
behavior won't be changed nevertheless, typically with the reasons why given
when the resolution is set to WONTFIX).
A ticket should not be marked INVALID simply because the issue reported is
required by a specification. A ticket should not be marked as invalid unless
the report is incorrect, or unless it does not report an issue in the software
against which it was filed.
Given that, having something both "RESOLVED" and "WONTFIX" is everything
working as-designed: there is no more work to be done, because an explicit decision has been made
to not change the behavior, even though the behavior is indeed wrong in some way.
The concrete meaning of "WONTFIX" is that all of the product's current developers have chosen not
to work on the issue. There *is* more work to be done on a "WONTFIX" issue, since a
"WONTFIX" issue is not resolved.
You seem to be using a different definition of the word "resolved" than that used in the
Bugzilla UI and insisting that everyone else use the same definition as you, even after it was
explained to you what the specific status "RESOLVED" means in terms of the Bugzilla
workflow. That's not very constructive, unfortunately...
A group of people sharing a system must agree on the definition of its symbols. We cannot
have a person consider that term foo means x, and another consider that it means y if x
and y can be contrary. We must use common definitions of statuses. If I am searching a
persisting issue in the ITS assuming that status RESOLVED means the issue is resolved,
but that the corresponding report was marked as RESOLVED by someone who used an
alternative definition which considers some unresolved issues "RESOLVED", there
is little chance for me to find the ticket. Therefore, I do think it is constructive to
ask that alternative definitions are prevented as much as possible.
The only place where it would have been "explained" to me what the "RESOLVED"
status means is in ticket #1285748: https://bugzilla.mozilla.org/show_bug.cgi?id=1285748
In comment #7, Byron Jones claimed "RESOLVED" would mean "not further action will be
taken here"[sic], specifying in comment #9 that this definition would be specific to computer
science. As dictionaries do not acknowledge such a definition, I rejected that in comment #10, and
no one has replied to that so far.
Dylan Hardison then suggested in comment #11 that "RESOLVED" would mean "Decide firmly on a
course of action". Before bothering to point out that this adjective would apply to humans, not to
tickets, I challenged Dylan in comment #13 to start by demonstrating that this definition would explain using
"RESOLVED" for the ticket where his comment was made, asking what would be that hypothetical course
Months later, there is still no answer to the challenges on both of these
suggestions. Considering that, and that these suggestions conflict with the
intuitive interpretation, I consider these suggestions wrong, as I explained in
ticket #1288913 comment #6.
Since BMO has been in place for more than a decade, I cannot believe some are suggesting that issue status
"RESOLVED" would not indicate a resolved issue. However, since you are the third person I see
proposing an alternative definition in a few months, and since the situation appears imperfect as you point
out with the case of "REOPENED", I just went to BMO and searched for a bug about an eventual issue
on the definition of "RESOLVED". There is no such report against either BMO or Bugzilla.
In terms of constructiveness, if the Status field is so bad that some contributor is
convinced that "RESOLVED" can mean something different that the expected
meaning, rather than proposing one more twisted definition, then he should:
1. Fix the issue
2. Until item 1 is completed, refrain from flagging reports of its consequences
as invalid even when these reports disagree with his own interpretation.
governance mailing list