Re: Ports & Packages infrastructure & feature import. (fwd)

[Jukka Ruohonen <jruohonen%iki.fi@localhost>]

On Wed, Aug 31, 2011 at 11:45:35PM +0400, Loganaden Velvindron wrote:
> Now if my current is slightly outdated compared to him due to some ``build
> issues'', the quality of my diff becomes shoddy.

I don't really see this as an issue, but your mileage may vary.

> If It's a device driver diff, I'd like users to test it in case it causes
> a regression of some other hardware revisions. It's hard to get
> testers to run current if they know that it often has issues as Mark
> previously said.

My guess would be that the people who are willing to test patches are
already running current. It is probably hard to get testers for hardware
diffs simply because of the small userbase.

But on the other hand, Linux is also quite prone to regressions, despite of
millions[1] of users who test things. If you think something like quirky
x86 hardware, maybe the only way to gain real test coverage is to have a
huge lab full of computers, like Microsoft.

Regressions happen, we all make them, and I would say this is related to the
fact that operating systems are probably the most complex software systems
known to man.

- Jukka.

[1] An arbitrary number.