Subject: Re: CVS commit: src/libexec/httpd
To: Hubert Feyrer <hubert@feyrer.de>
From: Bill Stouder-Studenmund <wrstuden@netbsd.org>
List: tech-userlevel
Date: 10/16/2007 17:47:29
--dc+cDN39EJAMEtIO
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Wed, Oct 17, 2007 at 01:46:06AM +0200, Hubert Feyrer wrote:
> On Tue, 16 Oct 2007, Jared D. McNeill wrote:
> >>>In your opinion, should we move more of NetBSD into pkgsrc? If so, how=
=20
> >>>do we handle maintenance of first party software in pkgsrc?
> >>
> >>If by "first party software" you mean what we have in base today (and=
=20
> >>what wasn't imported from random places), then I'd say: not at all.
> >
> >No, by "first party software" I mean the web server that was written and=
=20
> >imported (as I gathered from reading the rest of this thread) with the=
=20
> >intent of doing ongoing development and maintenance in the NetBSD tree.
> >
> >My question remains.
>=20
> I don't know, sorry.
> Last I've checked, NetBSD was an operating system project.

What exactly is an operating system?

For a lot of people, it's what we call NetBSD + pkgsrc.

> >>Have (keep/get) a small core system, and add the rest via a packaging=
=20
> >>system.
> >
> >So we should move more to pkgsrc? Where do you draw the line?
>=20
> Kernel, packaging tools, no idea what else.
>=20
> Where do *you* draw the line, what's next?

Personally? Not much more.

> Can I get Ruby on Rails with our shiny new webserver?

A) Well, how big would Ruby on Rails be? How many /bin/sh's? How many=20
/bin/ksh's?

B) What's its release cycle/life cycle like? One of the advantages I see
of pkgsrc is it has its own release cycle, and tends to release much more
often than NetBSD. If adding Ruby on Rails to base means we look out of
date really fast ('cause Ruby's been updated a lot), then we may not want
to do it.

C) How much of an enabler will Ruby be? Especially of more core-os-kinda=20
things? Like if someone had a cool feature we really really wanted in the=
=20
main system that was in Ruby, well, maybe we should do it.

My understanding of bozohttpd is that it is smaller than /bin/sh, so=20
(A) is less than one. Answer (B) is "slow." Not much is going on w/=20
bozohttpd. There may be some now that it's in NetBSD, but it's certainly=20
not like a lot of other projects. And for (C) I think it's an enabler of=20
other things people are doing now and want to do.

To be honest, if Ruby weighed in well on all these questions, the only=20
objection I'd have to adding it is that I prefer Python, so where's it.=20
:-) Though if we came up with a cool-enough-in-main-OS feature that really=
=20
wanted Ruby, I could cope with just Ruby coming in.

Take care,

Bill

--dc+cDN39EJAMEtIO
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (NetBSD)

iD8DBQFHFVuhWz+3JHUci9cRAqmcAJ43iaJyd60fN88EKnZ8RG0++javnACfVNVF
Q6lvV7Wbw8MzEQwIBU79Y40=
=m+4x
-----END PGP SIGNATURE-----

--dc+cDN39EJAMEtIO--