The Old Way Was Better

[Paul Robinson]

Gary W. Swearingen wrote:

>You'll recruit more testers by making releases and you'll recruit even
>more by naming the releases well.  What "well" is is the problem,
>because name choices have other effects too.
>

I agree except you will NOT recruit testers by making releases in this 
way. You will make enemies.

If something is named a -RELEASE, I and the rest of the planet expect it 
to be production ready code, capable of going into a live environment 
and the kind of software that I should be able to buy in a 
shrink-wrapped box.

If it is NOT production ready, it should be named as such. It should 
have a -BETA in the name. It should DEFINITELY not be named a -RELEASE. 
I know the amount of work that's gone into 5.x, and I know that there is 
a need for testers. What has happened with 5.x though is an absolute 
travesty. I know a lot of people will never, ever trust the FreeBSD 
release engineers again - they will refuse to run code released as a 
-RELEASE until they've heard it's safe. In effect, we've lost 
"customers". We've lost potential testers. We've lost new users, 
potential developers and people prepared to throw money at the project. 
We've lost a lot, because we -RELEASE'ed something before it was 
release-ready.

A policy on naming beta code as beta code is required.

Everybody here is aware of the fact that 5.x shouldn't be rolled out 
onto the payroll system just yet, but what does that say about the 
project as a whole when it's named as a -RELEASE, to somebody coming 
over to FreeBSD from the dark side?

>and were the old way sounds worse.  Unless there's a REAL old way
>where beta releases carry beta-type names.  As I indicated above, it's
>necessary to make a few beta releases and the only question is what
>they should have been named.
>

Keeping the betas named as betas would be fine. 5.0-BETA-1 should have 
been the name for 5.0-RELEASE. Then 5.0-BETA-2 for 5.1-RELEASE, 
5.0-BETA-3 for 5.2-RELEASE, then 5.0-BETA-4, 5.0-BETA-5, etc. then when 
the code is READY for a production environment and everybody agrees it 
rocks, we finally get to 5.0-RELEASE

Obviously, this can't be done on the 5.x branch now, but in the future...

-- 
Paul Robinson