1.7.1 release date?

Corinna Vinschen corinna-cygwin@cygwin.com
Sat Dec 5 09:39:00 GMT 2009


On Dec  5 00:46, Charles Wilson wrote:
> Maybe the following would work, as a transition scheme. It's a bit
> complex, but I think it avoids most of the problems folks have mentioned.
> 
> 1) current setup.ini continues to be associated with setup[-legacy].exe
> 
> 2) current setup-2.ini continues to be associated with setup-1.7.exe,
> even after setup-1.7.exe is renamed to setup.exe.
> 
> HOWEVER, the directory structure gets changed as follows:
> 
> release/    -> amphibius/
> release-2/  -> kiboko/
> 
> These names are taken from the five sub-species of hippopotamus:
> 
>     * H. a. amphibius (H. a. stands for "Hippopotamus amphibius", so
>                        this version has amphibius in its name twice)
>     * H. a. kiboko
>     * H. a. capensis
>     * H. a. tschadensis
>     * H. a. constrictus
> 
> upset is modified to generate "setup-kiboko.ini", and create a hardlink
> setup-2.ini.
> 
> One last time, upset is run on the release/ (ne' amphibius/) directory,
> to generate setup-amphibius.ini and a hardlink setup.ini with
> appropriate paths inside amphibius/.
> 
> "old" setup is called "setup-amphibius.exe" (with perhap hardlinks named
> "setup-legacy.exe" and "setup-win9xMe.exe") This newly compiled version
> of the old setup codebase would by default look for setup-amphibius.ini.
>  setup.ini is kept around only for those people that are using a
> previously-downloaded copy of old setup.exe.
> 
> "new" setup is called "setup-kiboko.exe" (with perhaps a hardlink called
> "setup.exe").  This newly compiled exe by default will look for
> setup-kiboku.ini.  setup-2.ini is kept around only for those people that
> are using a previously-downloaded copy of "new" setup-1.7.exe.
> 
> There exist, at the top level cygwin/ directory, the following (empty)
> files:
> 
> __CYGWIN_kiboko_is_CURRENT
> __CYGWIN_amphibius_is_win9xMe_only_UNSUPPORTED
> 
> 
> Note that in this scheme, kiboko would not simply be "1.7" by another
> name. It would also be 1.9.x, 1.11.x, up until we hit another SUPERMAJOR
> milestone (similar in magnitude to dropping support for an entire class
> of Windows operating systems, or the cygwin2.dll transition).
> 
> 
> The benefits of this scheme:
> 
> 1) people using the copies of setup.exe and setup-1.7.exe sitting on
> their harddrives right now, can continue to use them without surprises
> 
> 2) We avoid any confusion associated with "release-2" being current, but
> "release" being completely unmaintained, but with a name that implies
> active support.
> 
> 3) Hippo names are cool.
> 
> 4) It's extensible for later SUPERMAJOR milestones. In 15 years, we can
> worry about what comes after constrictus.

Sounds like a really good idea to me.

> Downside:
> 
> 1) Just as much mirror churn as the release-2/release <->
> release/release-legacy swap.  I'm not really convinced this is a
> terrible penalty.

I still don't like the idea to rename the directories, but I'm willing
to go along with it.  If the directories are renamed as you propose
here, we don't have a temporary collision between release and release-2,
so that's ok.

> 2) Folks might think that "kiboko" is a true "cygwin release" in the way
> Jaunty Jackalope or Feisty Fawn is.  In CVS terms, it's not a "tag" it's
> a "branch" -- a continuing line of development distinct from those that
> preceded it.
> 
> 3) Nothing intuitively says "amphibius" is old busted, "kiboko" is new
> hotness -- except the empty files I mentioned above, and perhaps text on
> a webpage somewhere...

Names like that have no inherent meaning and always need explanation.
Nobody outside of the Ubuntu community knows if Feisty Fawn is older
or newer than Jaunty Jackalope.  It doesn't matter, it's just a funny
name and the meaning is given extrinsically.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader          cygwin AT cygwin DOT com
Red Hat



More information about the Cygwin-developers mailing list