Date: Sat, 9 Sep 2000 16:16:34 +0200 From: Neil Blakey-Milner <nbm@mithrandr.moria.org> To: Steve Price <sprice@hiwaay.net> Cc: Will Andrews <will@physics.purdue.edu>, FreeBSD Ports <ports@FreeBSD.ORG> Subject: Re: PortsNG (was Re: Ports Options Paper) Message-ID: <20000909161633.A71013@mithrandr.moria.org> In-Reply-To: <20000909003743.B92984@bonsai.hiwaay.net>; from sprice@hiwaay.net on Sat, Sep 09, 2000 at 12:37:43AM -0500 References: <20000903052226.E1205@radon.gryphonsoft.com> <20000909003743.B92984@bonsai.hiwaay.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat 2000-09-09 (00:37), Steve Price wrote:
> Please let's forget about the 10 perl scripts, 3 acts of congress, and
> 2 acts of God for a few minutes. Let's define the problem fully first
> and then start talking about ways in which we can address them. I have
> a feeling that we'll come up with several distinct tasks that we can design
> together and code in parallel. For instance, we'll probably end up
> needing a better (read, more flexible) package format in the long run,
> but we must define our expectations of it before we jump in and start
> coding it otherwise we'll be back here next year doing the same thing.
> If we ever get through hacking hacks that is.
Firstly, I'm only responding to things that come directly to mind
immediately, and if I don't respond to something, you can be sure you've
got me thinking deeply. I'm pondering setting up a Wiki-like system to
describe what we all want, and how we each thinks it can be achieved,
and what background materials we make these assumptions from - OpenBSD
ports, NetBSD pkgsrc, Debian's dpkg/apt system, &c.
I don't think the package format is in the least significant to the
problem, except possibly the use of zip-like archives to only grab
headers, and to perform some sort of package signing. These are both
dealt with in the package format described and implemented in libh
currently.
> Sounds pretty easy so let's take this one step further. Let's suppose
> the same libfoo port is updated to a new version and now can also be
> built WITH_ICK. Here's where it gets tricky. Are WITH_BAR and WITH_ICK
> mutually exclusive or can both of them be on at the same time? If the
> former then the new system needs to have the smarts to recognise that
> there is a conflict and do 'something' about it. In the latter case
> we need to have in place the infrastructure to build libfoo, libfoo_bar,
> libfoo_ick, and libfoo_bar_ick.
One unexplored bit of functionality I had in my portconf dashboard a
year and a half ago required "multiple packages from one port", in that
it was an auto-explore on all the available options (that affected the
build) and generating a package with each set of compatible options:
{
{ !foo, !bar, !baz },
{ !foo, !bar, baz },
{ !foo, bar, !baz },
{ !foo, bar, baz },
{ foo, !bar, !baz },
...
}
I think this may be overkill, but it's probably something that can
easily be implemented in an automatic way, and definitely something that
can be implemented in a manual way (cf. OpenBSD flavours, portconf
'classes') even in the one-port-one-package way.
> Things like this are what we *must* nail down before we to decide to
> pilfer, purchase, or code it ourselves. Am I the only one that feels
> this way?
The way things usually work is:
a) You provide thoughts, and no code, and you're told you're just
talking hot air, and that you're an arm-chair general, and that you
don't understand all the implications;
b) You provide the code, but no documentation, and you're told that the
change requires documentation and a design document, and that you're
just hacking, and that without the design doc, you can't possibly
understand all the implications;
c) You provide code and documentation, and you get no feedback.
Neil
--
Neil Blakey-Milner
Sunesi Clinical Systems
nbm@mithrandr.moria.org
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000909161633.A71013>