Severity levels

[Ben Klein]

2009/5/4 Nicklas Börjesson <Nicklas.Borjesson at ws.se>:
> As I wrote in my earlier post, Austin told me about the voting functionality,
> and If that is considered when priorities are made, it is likely to keep
> things pretty on track, making my proposed changes far less important.
> I still think my thoughts aren't that off anyway, but now they feel a
> bit more "optional".

You're one guy against the world. So far, no one on this thread has
responded positively to your proposal to overhaul severities. I'd
suggest you stop acting like it's an inevitability

Once again bugzilla is a developer's tool, not a collection of data
for users. We already have the Wiki, forums and AppDB satisfying the
users' needs.

>>No, he's proposing to dump the developer-focused severity completely,
>>because "component + priority should be good enough", and replace it
>>with ill-defined, ambiguous Low, Medium, High, Critical. Blockers and
>>metabugs would also disappear under his proposed model, it seems.
>>After all, what good are metabugs to users? ;)
>
> Exactly. Ben's got it. :-)

Point is that metabugs, though useless to users, are important for the
REAL target audience of bugzilla: developers.

Repeat after me:
* Bugzilla is there for the developers, not the users
* Bugzilla's target audience is the developers, not the users
* Bugzilla's target demographic is the developers, not the users
* Bugzilla will fly up Tokyo Tower for the developers, not the users
* Bugs on bugzilla are examined, responded and fixed to by developers, not users
(OK, there are cases where that last one is false, but wherever code
is required ...)

A user-centric focus on bug priorities simply would not work with a
project as large (massive?) as Wine is.

> But blockers and metabugs wouldn't disappear.
> They would only lose their special classification.

Then they disappear. There would be no way to search for metabugs, for
example, whereas at the moment you can search for Blockers. There's no
point in keeping metabugs if there's no

> They would likely have it's priorities set to 1 by the developer reviewing.
> In what other way than they are highly prioritized are they different to any other bug?
> Something that must be fixed, must be fixed, regardless.

You're now implying that all bugs should be given equal priority. Some
bugs *can't* be fixed without limitations lifted in other areas (e.g.
introduction of Xinput2). They could still be severe (mouse-related
bugs could easily attain Major severity) but be a low priority due to
forces outside of Wine.

> To me, blocker is a class of bugs, not a level of severity.

Agreed, Blocker is a class of bugs, but severity is a neat way to keep
track of it. You can't have a minor or trivial blocker; all blockers
block development in their area.

> Mixing is up like is done now makes it:
> a) more complicated for users. b) more difficult to severity in statistics.

a) is not a consideration for bugzilla. It has to be easy for the
developers that respond.
b) is nonsensical. We're talking about two very different forms of
statistics. Bugzilla is the place for developer-side severity (which
is what's in place now); forums and AppDB are the places for user-side
severity (which is what you're suggesting). By definition, there is no
way to gather statistics on one when the other is used.

>>ill-defined
> I would go further than I'll-defined. I'd say non-defined.

Please quote with context! And having no definitions of the severity
levels is just asking for trouble. Surely you realise that every user
will have a different opinion on how serious their bug is, and what
the boundaries for "Low", "Medium", "High" and "Critical" are, which
will likely vary greatly from what the developers responding to the
bugs think? Conflict between developers and users is bad, and having
real definitions of the severity levels allows the devs to say "well,
your 'Critical' bug that has a simple but tedious workaround doesn't
fit the definition of Critical, so it's being downgraded to 'Minor'."

> The other things I talked about, drifting away from usability is a
> fairly rapid process(a few years) that I actually have experienced first hand
> (well second hand, actually), and it wasn't pretty.
> You joke about it, but the worst thing about it is that because it really only
> needs such a small skew to happen, it creeps up on you.
> Because "normal" get fixed far more often than "minor" bugs.

I can't tell what you're talking about here.
1) "A few years" is rapid?
2) "Drifting away" from usability when we've always (AFAIK) had the
current severity levels (but not necessarily the definitions), and
established that the system works well to assist developers
categorising the bugs?
3) "Such a small skew", as in handing over full control over what
priority should be given to the bug to the users? And before you say
it's not "full control", if it's not something that will seriously
influence the way bugs are prioritised, it's pointless to do such a
massive overhaul of the severity ratings.
4) "Normal" getting fixed more than "minor" is a problem?

>> Firefox and IE have drastically different success/failure/issues
>> when running in Wine, as do MS Word and
>> WordPerfect
>
> Yep, but I'd rather put I it like a Microsoft application has often other problems than externally developed applications(built-in vs using dll:s for everything).

Applications still have to be treated on a per-app basis. Every app is
different, and in extreme cases different versions of the same app use
violently different API calls.

> Anyway, looking at the forums now, games and 3d applications DO usually have different issues than normal desktop applications.
> It's more about controllers, DirectX and other stuff.

Correct, games and 3D applications do tend to use DirectX, whereas
office applications don't (tend to). Well done, astute observation. It
doesn't mean that a joystick fix for GTA:San Andreas will work with
Gunmetal, or that a WineD3D patch for COD4 will improve performance in
Supreme Commander. They're all individuals! (Chorus: Yes, they're all
individuals!) They're all different! (Chorus: Yes, they're all
different!) (I'm not) (shhh)

> And there are a LOT of posts. 30 new threads the last 24 hours. Lot's to wade through I you're only into D3D issues.
> If it was different lists, people could become a little bit more specialized.

That's what searching is for.