Re: Sets, subsets, syspkgs, and MK*

[Masao Uebayashi <uebayasi%tombi.co.jp@localhost>]

> So, let me get this right.  You're building your own binary distribution
> of NetBSD, and your own binary patches.  And still you want to inflict a
> lot of work to TNF and its members so that you can check that the binary
> patches you produced are bit-identical to the one TNF produced?

Not a lot of work if once done right.

> My opinion, as a member of TNF, is that TNF shouldn't care about other
> binary distributions of NetBSD,

I agree.

> and if I were to build my own binary
> distribution of NetBSD, I'd produce my own binary patches.

I'd use TNF's patches for libc.so.

I'd also produce binary patches for my local 3rd party programs exact same
way as TNF does (if build.sh once supports it).

> TNF has two kind of users:  binary users and source users.  It's
> perfectly acceptable to build binary patches for both, but why should
> TNF care about hypothetical hybrid users?

I'm not saying like that.

> Right now TNF isn't even properly taking care of its binary users, and
> you want to skip that and make TNF follow your own agenda?

I'm not saying like that.

> > > Flexibility is nice, but tying knots with your own limbs is just creepy.
> > 
> > Who said like that?
> 
> It seems to me that I am the one saying that.  I remember typing it, so
> it's no accident that it appeared in the mail.
> 
> > We all know binary distribution is restrictive.
> 
> Yes.  So why complicate things?

I'm not saying like that.  At least that's not my intention.

Do you think doing s/.if ${USE_YP} != "no// in lib/libc/Makefile & adjusting
set lists complicates things?

Masao

-- 
Masao Uebayashi / Tombi Inc. / Tel: +81-90-9141-4635