On Mon, 25 Oct 2004, John Cliff wrote:
I disagree. Our bug tracker is frightening. Lots of mysterious bugs which I don't know how they are possible at all. Lots of unconfirmed bugs. Lots of just plain very hard to fix bugs.
There are fewer crash bugs listed than we've had in the past, and most of those occur only in special situations (such as Win32-only),
I don't think Windows users will agree to be treated as a "special
situation" :)
Yes, win32 needs some love :)
and from the SF dl stats windows could be as much as 1/2 our userbase, Whatever the actual figure is, 10,000 dls is a lot to be considering a special situation. Admittedly tho, its kinda hard to fix windows bugs when theres virtually no one developing actively on it.
Of course what I meant was "Special situations like appearing only on a single platform." Although I'd also agree with the interpretation that uses 'special' in the 'special ed' sense, as regards Windows. ;-)
Also, I think my point about the character of the bugs was missed. Yes there are a lot of bug reports, and yes they're mysterious / hard / unconfirmed, but this is an improvement over past releases where we've had overt showstopper crash bugs that afflicted everyone. My point is that the ratio of bugs that affect everyone all the time to bugs that occur only sometimes on some platforms has noticeably improved from the past, which means we're starting from a better point than before. That's a good thing. Doesn't mean that the listed bugs don't need to be fixed.
Anyway, my question about how to measure when we've achieved our bug fixing goal for this release has not received any answer. How do we determine when we've fixed enough bugs, or the correct set of bugs, to be able to do a release? It is easier to get folks motivated to do bug fixing if there is a tangible goal we can focus on.
Bryce