Gentoo Archives: gentoo-scm

From: Fabian Groffen <grobian@g.o>
To: gentoo-scm@l.g.o
Subject: Re: [gentoo-scm] Thin Manifests for git
Date: Mon, 19 Apr 2010 09:45:10
Message-Id: 20100419094728.GA3789@gentoo.org
In Reply to: Re: [gentoo-scm] Thin Manifests for git by Maciej Mrozowski
On 19-04-2010 11:38:30 +0200, Maciej Mrozowski wrote:
> On Monday 19 of April 2010 08:21:27 Fabian Groffen wrote: > > On 18-04-2010 23:52:34 +0200, Michał Górny wrote: > > > I've did as much as I could without getting more information, and > > > turning everything around. What it can is: > > > 1) Detect '.git' and enable Thin Manifests (only dirty check), > > > > how about we use a file like manifest1_obsolete to switch from fat to > > thin? > > Any reason for this? > > It shouldn't cause any merge conflicts as the difference between 'fat' and > 'slim' are lines added/removed (I suppose DIST hashes would be generated in > the same form as they are now, so SHA1, SHA256 and RMD160). It's just over- > complicating things imho. > > Portage already is aware of scm vs rsync repositories as for instance some > repoman checks are omitted for those scm ones. Therefore it seems correct to > rely on this mechanism and let portage/repoman generate right (TM) Manifest > for repository being used.
In my opinion they are dirty hacks that rely on their consumers: cvs (Gentoo) and svn (Gentoo Prefix) are treated as full, while Git (Funtoo), hg and bzr (Fauli) omit e.g. the ChangeLog check. -- Fabian Groffen Gentoo on a different level