Re: [HEADSUP] Let's start a Cygwin 1.7 release area

[Christopher Faylor <cgf-use-the-mailinglist-please@cygwin.com>]

On Sun, Apr 20, 2008 at 02:46:28PM -0400, Charles Wilson wrote:
> Christopher Faylor wrote:
>> Since most of what I did was automated, I probably ended up deleting too
>> much.  Please take a look at the package directories that you maintain
>> and verify that the release-2 directory contains something that makes
>> sense.
>
> If we find a problem, how should we fix it? For instance, release(1) 
> inetutils has prev: 1.3.2-37, curr: 1.3.2-40, and test: 1.5-3. release-2 
> has only 1.5-3, but its setup still specified that 1.5-3 is test (it also 
> specifies the non-existant 1.3.2-40 and 1.3.2-37 packages as curr: and 
> prev:, respectively).
>
> I'm not ready to make inetutils-1.5-3 curr: on the release(1) side. 
> However, since ALL of release-2 is test:, I could see doing any of the 
> following:
>
> (1) copy over the 1.3.2-40 packages, and remove the prev: entry from 
> setup.hint
>
> (2) change the release-2 setup.hint so that on "that side", inetutils-1.5-3 
> becomes curr: (and remove all references in setup.hint to the "missing" 
> packages).
>
> (3) change the release-2 setup.hint by removing all references to the 
> "missing" packages, but otherwise leave things along. That is, on the 
> release-2 side, "inetutils" would ONLY be available as a test release, 
> until I manually promote 1.5-3.
>
> But I'm not really sure how union fs works. Can I edit setup.hint in place? 
> Can I "copy" back the "missing" files, or is there some special procedure 
> that makes them "re-appear"?

I think the simplest thing to do is remove the test from setup.hint.  It's
all test anyway.  Anyone who is using this should be extremely aware of
the fact that it's unstable.

Apparently there is a bug in this version of unionfs, though.  If you
try to recreate a missing file by just copying it in or touching it you
get a "File exists" error and that's not right.  You aren't supposed to
have to know that there is something special going on and a mysterious
error like that sort of breaks that assumption.

cgf