Re: I think binary-branch is almost ready for play time

[Bruce Korb]

"Gary V. Vaughan" wrote:
> 
> Hi Bruce,
> 
> This is how I envision the logistics of integrating your branch to avoid CVS
> merge hell:
> 
>         * Finish the MLB->HEAD merge.
>         * Flush the patch queue (for HEAD and branch-1-4)
>         * Release libtool-1.4.1 (libtool-1.4-p1?) from branch-1-4
>         * Release 1.5b from HEAD, and create branch-1-5
>         * integrate your patch into HEAD (now 1.6a)
> 
> This relies on the MLB merge going well, and assumes that MLB is in good
> enough shape that only a few monthly betas will be needed before 1.5 is a
> resonable prospect.  I still haven't had time to play with MLB, but it seems
> to be widely used and probably is fairly stable by now, I guess.
> 
> Thoughts?

That's fine.  I think I would be inclined to leave the binary-branch
alone until the merging is done.  I *really* don't want the emitted
shell text to go too far afield from the main/mlb branch.  The steps
as I see it are (once bb is in sync with main):

1.  make a change that accomplishes a small goal:
    o fixes a bug
    o unifies common code into global macros
    o moves towards a real program (as opposed to a shell
      script emitter wrapped around the shell script).

2.  emit ltmain.in and verify that either the program will behave
    identically to before, or the bug is fixed.

3.  check in the generated ltmain.in into the main line

4.  Repeat until done.

That means that each step #1 should be small and safe.
Thoughts?  :-)