Date: Mon, 15 Nov 2004 22:10:21 -0600
From: "Karl O. Pinc" <kop@meme.com>
To: Miles Bader <miles@gnu.org>
Subject: Re: Tagline tagging failings -- Was: Re: [Gnu-arch-users] tla mv gets
	me an error next commit
Message-ID: <20041115221021.Y16278@mofo.meme.com>
References: <20041115151313.I8700@mofo.meme.com>
	<41995266.6060400@arbash-meinel.com>
	<20041115195312.I16278@mofo.meme.com>
	<buod5yeef5n.fsf@mctpc71.ucom.lsi.nec.co.jp>
	<20041115201538.O16278@mofo.meme.com>
	<buopt2ecz9b.fsf@mctpc71.ucom.lsi.nec.co.jp>
	<20041115205438.U16278@mofo.meme.com>
	<buois86cwx9.fsf@mctpc71.ucom.lsi.nec.co.jp>
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
In-Reply-To: <buois86cwx9.fsf@mctpc71.ucom.lsi.nec.co.jp>;
	from miles@lsi.nec.co.jp on Mon, Nov 15, 2004 at 21:09:06 -0600
Cc: gnu-arch-users@gnu.org, "Karl O. Pinc" <kop@meme.com>
Precedence: list
Sender: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Errors-To: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Status: RO
Content-Length: 3598
Lines: 73

On 2004.11.15 21:09 Miles Bader wrote:
> "Karl O. Pinc" <kop@meme.com> writes:
> >> As I described in my earlier message, it seems perfectly possible
> to
> >> do the job efficiently without any such "keeping records", and
> >> desirable to do so, as such record-keeping has a number of fairly
> >> severe problems[*].
> >
> > It's not like I know what I'm talking about here, but I've doubts
> > about your statement.  Because if it's _really_ possible to do the
> > job without "keeping records" then there would be no need for
> > explicit tagging _ever_, and arch wouldn't have it.
> 
> No, that's incorrect.
> 
> Explicit tags are _exactly_ like taglines:  they describe an absolute
> piece of state in the source tree, a file-name-to-id association.

Explicit tags serve exactly the same _function_ as taglines, but they
are a _different_ 'recordkeeping' implimentation.  And my point is that
they are not interchangeable.  In some cases explicit tags are required.
So, why not make other sort of in-tree recordkeeping required to
support some functionality?

> "Keeping track" of mv/rm/add would _not_ be "absolute" -- it records a
> _change_.  Recording changes to a dynamic source tree is problematic
> your records only makes sense in a particular context, and so you must
> be very careful to make sure it is kept up to date with respect to any
> further changes -- some of which may happen outside your control.

I understand. And the advantage of taglines is that you can use any
tool to manipulate the tree.  But in those cases when you still need
explicit tagging you're stuck and limited to using the
revision control tools.  
> > I'm not impressed with taglines, for reasons given in a previous
> email.
> > I'm not asking everybody to give up taglines, I'm asking everybody
> > who uses taglines to give up on fixing the partial commit problem.
> 
> Ok; now that I've explained why you're wrong, can you please stop
> asking
> this?

I could be wrong to speculate that explicit tagging makes certain
kinds of partial commits easier. (And so it'd be eaiser to
limit partial commit support to those using explicit tagging,
which is where this discussion started.)  But that is not what you've
been argueing against.  You seemd to turn to arguing against any sort of
in-tree recordkeeping,  Which struck me as impossible, and
I could not resist saying so.  No doubt this was a mistake
and I'm sorry, because surely this has been discussed before.

In any case I think we understand each
other and I'm willing and ready to stop.  You clearly have a
clue when it comes to the existing tla implimentation and
may very well be able to show that there's an easier way
to support partial commits than to use a recordkeepding
technique in the tree similar to explicit taggings.  You may
have done so already in your very first post which mentioned
implimentation details, before we got to explaining what
we both know to each other.  I didn't mean to troll and
have you respond to me, which sidetracked the discussion
away from why your suggested implimentation is simpler
than recordkeeping in the tree.  But there's no point in
continuing down that road either as I'm not in a position
to understand the merits of your suggestion.

So yes.  Let's stop.

Karl <kop@meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein


_______________________________________________
Gnu-arch-users mailing list
Gnu-arch-users@gnu.org
http://lists.gnu.org/mailman/listinfo/gnu-arch-users

GNU arch home page:
http://savannah.gnu.org/projects/gnu-arch/