From: David Bremner <david@tethera.net>
To: Jani Nikula <jani@nikula.org>, notmuch@notmuchmail.org
Subject: Re: [PATCH] RFC: impliment gzipped output for notmuch dump
In-Reply-To: <87wqfdbblz.fsf@nikula.org>
References: <874n2kipq4.fsf@qmul.ac.uk>
	<1396056046-2247-1-git-send-email-david@tethera.net>
	<87wqfdbblz.fsf@nikula.org>
User-Agent: Notmuch/0.17+150~g2743557 (http://notmuchmail.org) Emacs/24.3.1
	(x86_64-pc-linux-gnu)
Date: Sat, 29 Mar 2014 09:29:15 -0300
Message-ID: <87ob0pdw84.fsf@zancas.localnet>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: list

Jani Nikula <jani@nikula.org> writes:

> Is it reasonable to require zlib? It's possible to have conditional
> build for this, but it will be laborous to do it cleanly. I could be
> persuaded either way, perhaps more easily to just requiring it.

If I recall correctly, xapian needs zlib.  At least on Debian this is a
hard dependency. What is needed additionally is the zlib headers.

> If you do require zlib, the missing dependencies error message from
> configure should include a check for have_zlib and a note about it
> similar to other required dependencies.

good point.

> Also, it would seem natural to support gzipped input in restore as
> well... do you think it would be silly to support gzipped output only
> internally until we've added gzipped input too? This could be in a later
> release in the future.

I couldn't decide which was sillier: to have have gzipped output but not
input, or to leave off the command line argument (it's something like 3
lines of diff). I lean to making it accessible from the CLI to better
support testing.

d