Re: Sets, subsets, syspkgs, and MK*

["Aaron J. Grier" <agrier%poofygoof.com@localhost>]

On Tue, Dec 15, 2009 at 10:55:44PM +0900, Masao Uebayashi wrote:
> My point is to realize that most USE_* variables change signature
> globally.  USE_YP even changes libc signature.

- contents of library files and list of files in distribution are both
  affected by USE and MK variables.

> We should audit all those use and make them modular == no signature
> change but be "subsets".

I think you are looking for general case solution.  for first draft of
binary patches, assume that library signatures do not change.  for
second draft, create a method to determine MK variables from individual
library.

making these "subsets" is very intrusive to a lot of code, and does not
allow size savings.  code can be moved to libYP, but in order to keep
libc same between MKYP=yes and MKYP=no, libc must continue to have all
YP code in it.

> I was confused because those are "almost done", but not finished.  Most
> opinions I've got from people is "why do you do that"?  No one has answered
> my question - why don't we distribute binary patches?

base.tgz is an ugly form of binary patch, yes?  ^.^

NetBSD project does not distribute binary patches because nobody has
created automated method to do so.

-- 
  Aaron J. Grier | "Not your ordinary poofy goof." | 
agrier%poofygoof.com@localhost