From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <ddb5e5af66899d634068e1c409190150@quanstro.net>
From: erik quanstrom <quanstro@quanstro.net>
Date: Mon,  9 Mar 2009 11:11:16 -0400
To: 9fans@9fans.net
In-Reply-To: <df49a7370903090755p879ac84k1065d404cafafd43@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Subject: Re: [9fans] vac flattens trees?
Topicbox-Message-UUID: b6abf908-ead4-11e9-9d60-3106f5b1d025

> i suppose your preferred behaviour would be
> for it to do a treewise merge of the paths and
> vac the shallowest root that contains all the
> paths, missing out all contents that aren't
> under one of the named paths.
>
> this seems like quite an involved operation,
> when you can get much the same result
> by building a custom namespace containing the
> trees you want (admittedly, a recursive mntgen
> would be useful here).

wouldn't work on p9p.

it's hard to comment on the current behavior without
knowing the reasoning.  given your use case, it seems
like tar's model makes more sense.

a straightforward solution would be to have vac take
a proto.  i would think the -e flag would not scale.
then you can give a single root and include only the
stuff you want without changing current behavior.
and the bonus is that it works a lot like mkfs.

- erik