Mailing-List: contact gump-help@jakarta.apache.org; run by ezmlm
Message-ID: <145701c3b9b1$54e8aab0$9c27fea9@tsws1>
From: "Adam R. B. Jack" <ajack@trysybase.com>
To: <gump@jakarta.apache.org>
References: <bqke74$636$1@sea.gmane.org>
Subject: Re: Gump builds a-la-carte?
Date: Wed, 3 Dec 2003 08:22:51 -0700
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit

I think ideas like this are very interesting, I love this out-of-the-box
type thinking separating Gump past from Gump future. With the Python
codebase such things become much more possible. I say we handle issues with
"monster dependencies" (such as with Ant) either by a flag

Not the same, but a different approach is that when Gump does it's module
updates it checks for 'activity'. CVS (in quiet mode) only list updates, and
is silent if nothing changed. As such, a flag is set as to wether the module
(not project, somewhat crude). I document this right now (to try to display
when something is stale) but it could be used as an optimization. Gump has
mutliple code paths these days, and optimization switches, this sort of
stuff is very doable.

Note, with RSS the "nagged to death" issue isn't such a problem. Folk can
use their RSS reader to tap into a workspace (a Stefan, or Gumpmeister), a
module (e.g. Ant), or a project & they can override the 'update window' to
hours/mins/days, whatever they want. Basically, with RSS output we can nag
up the wazoo w/o being a pain. [I am just tweaking the RSS feed, a couple of
presentation bug, I'll post here when ready to be viewed. I also want some
help from Blogmeisters on if Gump ping topic serves, etc.]

There is so much *good stuff* we can do with this start. We need more active
coders in Gump, we (I) need to document what is there (the Python code is
getting pretty manageable now), and we need to recruit.Gump is a social
experiment that needs more close friends. Thoughts on how?

regards,

Adam