merge reviews

[Bill Nottingham]

Jeff Garzik (jgarzik at pobox.com) said: 
> After a large survey, it is readily apparent that many of these 242
> have been untouched for -years-, for packages that have been merged
> into Fedora and used happily for -years-.
> 
> Further hundreds of other reviews outside your 242 are listed as
> assigned to a reviewer, but making no progress after multiple years.
> 
> These merges are crowding out new packages
> that need merging, on bugzilla's list of "packages that need a review."

That logically does not follow. If they're not being touched,
and not making progress, then there's obviously no review resources
being spent on them, which means they're not taking anything away from
the review resources for other new packages. (We're still having a
good number of new packages approved each week.)

> As such, getting any package into Fedora requires navigating an
> informal, ever-changing process, where the chief attributes for
> success involve (a) knowing a Red Hat employee or (b) public,
> sometimes repeated begging on fedora devel.

... how is it ever-changing? The review and sponsorship process has been
relatively static for a long time. Yes, sometimes you have to request
reviewers, or swap reviews, but that's been the case for quite a while.

> The proper course of action for already-merged packages is to file
> bugs against those packages.  We have an entire -team- of people
> looking at kernel bugs, while this silly "merge review: kernel" sits,
> ignored, for several years.

Except you'll need to have some sort of tracking of packages
which haven't yet been processed looking for these bugs, so people know
what packages to look at first. Which then ends up looking an awful lot
like the current merge review list.

> The current package review system is failing miserably at separating
> wheat from chaff, is very chaotic, and non-deterministic.  "Merge
> review: kernel" is pure noise.

I'd like not to assume the worst, but given your mass closing of some
review bugs, plus your arguments here about why, plus your request for
a review swap earlier, I'm having trouble reading this as anything other
than a transparent frustration at your package not getting reviewed
fast enough for your liking, with an unsaid assertion that it's part of
the 'wheat' above. 

Right now, we have a dearth of review resources. This leads to both
merge reviews having no activity, and new package reviews sometimes having
no activity. However, if a new package is important enough, someone's going
to have enough self-interest to pick up the review, or enough self-interest
as maintainer to swap reviews for someone else in the same boat. Frankly,
that's the sort of intrinsic motivation common to open source... assuming
differently, that there would or should be some sort of nebulous 'review
community' sitting there waiting to take on a bunch of new submissions from
the mountain seems awfully unrealistic.

Bill