Re: Release plans

Hi, another week and Sunday gone so time for simple numbers...

340 regressions <-- release announcement
356 regressions <-- release announcement + 1week
339 regressions <-- release announcement + 2weeks(rc1)
322 regressions <-- release announcement + 3weeks(rc2)
325 regressions <-- release announcement + 4weeks
326 regressions <-- release announcement + 5weeks(rc3)

List of fixed bugs on rc3 release was long and impressive. Unfortunately
for regressions there is a stagnation.

Release plans

[Michael Stefaniuc]

On 06/13/2010 10:39 AM, [e-mail] wrote:
> Hi, another week and Sunday gone so time for simple numbers...
>
> 340 regressions<-- release announcement
> 356 regressions<-- release announcement + 1week
> 339 regressions<-- release announcement + 2weeks(rc1)
> 322 regressions<-- release announcement + 3weeks(rc2)
> 325 regressions<-- release announcement + 4weeks
> 326 regressions<-- release announcement + 5weeks(rc3)
>
> List of fixed bugs on rc3 release was long and impressive. Unfortunately
> for regressions there is a stagnation.

Regression bugs do get fixed too but "new" regressions are added at the 
same rate. A lot of the of the newly reported regressions happened 
before 1.2-rc1; it looks like people are using the release candidates to 
test their favorite app with and report the regression they find. And 
that is good; just screws up the statistics ;)

bye
    michael

Release plans

[James McKenzie]

Michael Stefaniuc wrote:
> On 06/13/2010 10:39 AM, [e-mail] wrote:
> > Hi, another week and Sunday gone so time for simple numbers...
> >
> > 340 regressions<-- release announcement
> > 356 regressions<-- release announcement + 1week
> > 339 regressions<-- release announcement + 2weeks(rc1)
> > 322 regressions<-- release announcement + 3weeks(rc2)
> > 325 regressions<-- release announcement + 4weeks
> > 326 regressions<-- release announcement + 5weeks(rc3)
> >
> > List of fixed bugs on rc3 release was long and impressive. Unfortunately
> > for regressions there is a stagnation.
>
> Regression bugs do get fixed too but "new" regressions are added at 
> the same rate. A lot of the of the newly reported regressions happened 
> before 1.2-rc1; it looks like people are using the release candidates 
> to test their favorite app with and report the regression they find. 
> And that is good; just screws up the statistics ;)

I'll second this as there have been "My application works with Wine 
1.0.1, but does not with Wine-1.2-rc(x)' messages in Wine Users.  Of 
course, this leads to bug reports.

I'll start looking at regressions as soon as I can.

James McKenzie

Release plans

Hi, another week and Sunday gone, but this time i tried to look a bit
closely to the numbers...

340 regressions <-- release announcement
356 regressions <-- release announcement + 1week
339 regressions <-- release announcement + 2weeks(rc1)
322 regressions <-- release announcement + 3weeks(rc2)
325 regressions <-- release announcement + 4weeks
326 regressions <-- release announcement + 5weeks(rc3)
325 regressions <-- release announcement + 6weeks(rc4)

3rd week in a row and unfortunately these numbers don't change significantly
to zero. Based on these following numbers we can't even say, that we
close at the same rate as new are opened. Closer look to "fixing" capacity
or in other words what is behind -1 fixed regression for this week:

+24 Newly marked, opened (new/unconfirmed)
-16 Closed, resolved fixed
-09 Not a regression (invalid, duplicate)
====
-1

More IN than OUT could show that stable release is far a way.

Another relese barometr could be Milestones 1.2:

59 bugs <-- release announcement + 5weeks(rc3)
46 bugs <-- release announcement + 6weeks(rc4)

Decoding shows much better numbers than in case of regressions:
+02 Newly marked, opened (new/unconfirmed)
-11 Closed, resolved fixed
-04 Not a Milestone-1.2
====
-13

W.

Release plans

[Scott Ritchie]

On 06/20/2010 01:45 PM, [e-mail] wrote:
> Hi, another week and Sunday gone, but this time i tried to look a bit
> closely to the numbers...
>
> 340 regressions <-- release announcement
> 356 regressions <-- release announcement + 1week
> 339 regressions <-- release announcement + 2weeks(rc1)
> 322 regressions <-- release announcement + 3weeks(rc2)
> 325 regressions <-- release announcement + 4weeks
> 326 regressions <-- release announcement + 5weeks(rc3)
> 325 regressions <-- release announcement + 6weeks(rc4)
>
> 3rd week in a row and unfortunately these numbers don't change significantly
> to zero. Based on these following numbers we can't even say, that we
> close at the same rate as new are opened. Closer look to "fixing" capacity
> or in other words what is behind -1 fixed regression for this week:
>
> +24 Newly marked, opened (new/unconfirmed)
> -16 Closed, resolved fixed
> -09 Not a regression (invalid, duplicate)
>
> -1
>
> More IN than OUT could show that stable release is far a way.

I'll note that you'll see a very similar measure when just looking at
Wine bugs in general.  We are, nevertheless, making progress ;)

Still, I do support delaying the release until it feels like there's a
substantial drop in non-deferred patches.  That's the sign that tells us
we've run out of easy enough release bugs/regressions to fix and may as
well release.

Thanks,
Scott Ritchie

Release plans

Hi, another week and Sunday gone. Lastly Scott had a nice idea about
release & "drop in non-deferred patches". I like it too, but no more
stat in this area ;)

340 regressions <-- release announcement
356 regressions <-- release announcement + 1week
339 regressions <-- release announcement + 2weeks(rc1)
322 regressions <-- release announcement + 3weeks(rc2)
325 regressions <-- release announcement + 4weeks
326 regressions <-- release announcement + 5weeks(rc3)
325 regressions <-- release announcement + 6weeks(rc4)
308 regressions <-- release announcement + 7weeks(rc5)

This nice drop consist of:

+04 Newly marked, opened (new/unconfirmed)
-13 Closed, resolved fixed
-08 Not a regression (invalid, duplicate)
====
-17

But it looks like, we can't have nice number in both regressions and
milestones.

59 bugs <-- release announcement + 5weeks(rc3)
46 bugs <-- release announcement + 6weeks(rc4)
43 bugs <-- release announcement + 7weeks(rc5)

Closer look reveals only one was fixed this week:

+00 Newly marked, opened (new/unconfirmed)
-01 Closed, resolved fixed
-02 Not a Milestone-1.2, back off Milestone-1.2
====
-03

W.