Re: Getting serious about releasing

[Luis Villa <louie ximian com>]

Some meta-commentary only here; I think discussion about specific bugs
belongs in the bugs.

On Tue, 2002-04-23 at 20:19, Sander Vesik wrote:
> > > > 75150: automatic updating of menus doesn't work in panel 
> > > IMHO puntable, just leave the impression there is no such thing as
> > > auto-updating. Then re-add in 2.0.1 or whenever.
> > 
> > Another one where we can punt by disabling the feature.
> >  
> 
> I think we need three lists - "disable feature", "puntable" "non-puntable"
> - not all punting needs an action to be taken.

I think it'll be fairly clear (and probably discussable on a
case-by-case basis anyway) which bugs are solvable by disabling an
option/feature/button and which are not; having additional lists and
keyword merely makes figuring out what to do even more complex.
Additional administrative/organizational complexity is not what we need
right now; we're already seeing bugs with 6-10 keywords on them. But I'm
strongly agreed that it is something for maintainers to keep in mind as
an option, and for me when I'm whining at them to 'just fix it'. :)

> > > > 75232: gtkhtml view in Nautilus crash
> > > > 75387: Nautilus crash pressing escape while icon browser is loading
> > > > 75120: Nautilus crash on startup (seems to be fixed, should close bug)
> > > 
> > > crashes aren't really puntable.
> > 
> > Yeah they are, if they are sufficiently obscure. No way we're going to
> > ship with no known crashes.
> > 
> 
> we should at the very leats strive to ship with no known crashes, beacuse
> according to Murphy, teher will otherwise be a gang of mad vocal gnomes
> somewhere who practice this as their daily obcsure thing to do. 

I don't necessarily agree with Sander's rationale ;) but I agree that
any trivially repeatable crash should not be considered puntable until
very late in the cycle. A big linux/GNOME selling point always has been
stability; we should be working as hard as possible not to compromise
that. [Note that heisenbugs that aren't reproducible or whatever don't
fall in this category.

> > > I snipped the individual ones, but "help" not working is not puntable. If
> > > we don't have help for a spcific thing, then we should display a dialog
> > > saying so. clicking on help needs to give an intelligent response.
> > 
> > If someone patches it sure, if they don't I'm willing to ship the
> > GWeather properties dialog with broken help. We have much worse
> > problems. e.g. GWeather parses the Yahoo weather site HTML and breaks
> > every time Yahoo gets a cosmetic overhaul. ;-)
> > 
> > Of course maybe these should all be under "not puntable but can be
> > made puntable by removing the help button"
> > 
> 
> Possibly. or if there was a stock "no help exists for this, please help
> the docs project" dialog or similar it could go theer instead of
> disabling. or something similar.

Again, agreed with Sander- help not working isn't puntable. Either the
button must go or some help or text must be there. That's just an
unacceptable newbie trap, especially if we leave enough other bugs in
that people are going to be hitting help quite a bit :)

FWIW, there are now only 24 open bugs from Sander and Havoc's combined
can't punt lists, not counting the help and crash issues. We've still
got a ways to go but it is definitely a very surmountable hill.
Luis