What's still needed to build the Windows 2.0.1 installers?

[Ian Clatworthy]

John Arbash Meinel wrote:

> Time, and lots of patience....
> 
>   bzr 2.0.1-1 has finally been built.

Thanks.

> It has now taken around 10 full builds before I feel like I've gotten
> stuff sorted out for 2.0.1-1... We really need to have this be smoother,
> or I'm going to have to step down and ask someone else to do this.
> 

>   bzr-explorer	? => 0.8.3

0.8.2 was in 2.0.0.

> Note that I do *not* have 2.1.0b1 built yet. But the expected changes are:
>   bzr		2.0.0 => 2.1.0b1
>   bzrtools 	2.0.0 => 2.1.0b1
>   qbzr 		0.14.2 => 0.15
>   bzr-svn 	1.0.0 => something compatible with 2.1.0b1
>   subvertpy 	0.6.9 => 0.7.? (0.7.0 fails to build)
>   tortoisebzr 	no change (0.3.0, 0.3.1 needs updated bzr)
>   bzr-explorer 	=> 0.8.3
>   bzr-rewrite 	no change (0.5.4)
> 
> The 2 main blockers are:
>   1) Needing a newer bzr-svn and subvertpy
>   2) Out of steam for fighting with issues (thus not even attempting
>      until (1) is solved.)

So I really think we need to change our process in a few ways to restore
some sort of Release Manager sanity here:

1. patches which change the minimum API in bzrlib need to be announced
   WHEN THEY LAND so plugin developers have time to respond

2. Plugin developers should become repsonsible for tweaking the version
   number released in the Windows installer.

Would that help, assuming we have 2 branches for bzr-windows-installers,
one for 2.0.x and another for 2.1.0betaX?

Ian C.