Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 19 Jan 2010 16:35:46 -0500
From:      Wesley Shields <wxs@FreeBSD.org>
To:        Doug Barton <dougb@FreeBSD.org>
Cc:        "Philip M. Gollucci" <pgollucci@p6m7g8.com>, freebsd-ports@freebsd.org
Subject:   Re: Why change to plist-sub for pkg-message?
Message-ID:  <20100119213546.GD54096@atarininja.org>
In-Reply-To: <4B550C9E.6040303@FreeBSD.org>
References:  <4B54D17E.8000508@FreeBSD.org> <4B550975.8090800@p6m7g8.com> <4B550C9E.6040303@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jan 18, 2010 at 05:36:30PM -0800, Doug Barton wrote:
> On 01/18/10 17:23, Philip M. Gollucci wrote:
> > That particular one is questionable, but I'm sure if every post-install
> > message is in pkg-message or files/pkg-message.in it will help with
> > 
> > 
> > http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/122877
> 
> Ok, now I see where you're going with this. :)  With respect to Wesley,
> the approach in the patch is bass-ackwards. You don't want to do
> anything with files named *pkg*message*, there are just too many
> variations. The only thing you want to worry about is
> /var/db/pkg/*/+DISPLAY. That way you can be sure of several useful
> things ... it's persistent (I.e., users can bring it up again with
> pkg_info), it's already post-processed, etc. etc.

Wow, that's a patch from the past. ;)

You are correct that there are too many variations, not to mention my
patch failed to handle the case where a port is installed as an
EXTRACT_DEPENDS. That is part of the reason that the PR is in limbo - it
clearly needs more work. I would love to find time to clean it up
because I still think the feature is useful. Thanks for the hint on
using +DISPLAY for this. If I can ever find the time to re-work the
patch I will certainly explore that option.

> Modesty aside portmaster has some pretty mature code to deal with this,
> I suggest giving it a look before trying to reinvent it.

I love the fact that portmaster does this and will look at the code if I
can find time to re-work the patch. If anyone wants to get to it first
then please do. I'm willing to give a review and pointers to some of the
pitfalls I've noticed with my original approach.

> As for the rest of the patch, I appreciate the POLA-awareness, but I
> would make this opt-OUT instead of opt-in, tool authors can easily opt
> out of it with a one-line fix, whereas the feature is very useful for
> the average user and I believe it should be enabled by default. It's one
> of the things that people tell me most often that they like about
> portmaster.

Point taken. I chose to not break POLA but can see your viewpoint also.
As is often the case with these situations I'm agnostic and leave it up
to the powers-that-be (portmgr) to decide on that.

-- WXS



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100119213546.GD54096>