Call for testers for yet another ports upgrade program, ports+

[Yoshihiro Ota]

On Sun, 29 Jul 2007 02:35:30 -0500
"Matthew D. Fuller" <fullermd at over-yonder.net> wrote:

> On Sat, Jul 28, 2007 at 10:14:28PM -0400 I heard the voice of
> Yoshihiro Ota, and lo! it spake thus:
> > "Matthew D. Fuller" <fullermd at over-yonder.net> wrote:
> > > 
> > > No, it's not.  Lots of ports change dependancies based on what's
> > > installed, so if you install one thing you have to assume that any
> > > other thing you intend to build in the future may change its list
> > > because of that.  For a trivial demonstration, look at how many
> > > use the ".if ${HAVE_GNOME:Msomething}" pattern.
> > 
> > It doesn't matter how many you have.  There are only finite number
> > of ports and configurations.  You will finish configuring at one
> > point.
> 
> No, you mistake me.  I'm not talking about ports _configuring_.  I'm
> talking about ports _installing_.
> 
> You can recursively configure all the ports, but the dependancies can
> still change AFTER that, when other ports get INSTALLED.  As soon as
> you install one port in your process, every other port you're dealing
> with can potentially change its dependancies, with NO actual user
> interaction or configuration.


I don't think I understand very well on this issue.  I have a couple
of questions.

1. Is this about RUN_DEPENDS in makefiles?

2. Is this about DEPENDS in makefiles?

3. Or, is this about the "configure" scripts detect extra
libraries at run time?

4. If so, does the ports system keep track of such information?
At least, I couldn't find anything alarming of this in the makefiles.
Is that something the pkg_* tools detect?

5. Do you know how portupgrade or portmaster extract such dependencies?

6. Can you give me an easy example of such port or scenario?

Thanks,
Hiro