From owner-freebsd-questions Thu Jul 24 11:30:22 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA16786 for questions-outgoing; Thu, 24 Jul 1997 11:30:22 -0700 (PDT) Received: from intelsat1.intelsat.int (intelsat1.intelsat.int [164.86.100.3]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA16778 for ; Thu, 24 Jul 1997 11:30:19 -0700 (PDT) Received: (from smap@localhost) by intelsat1.intelsat.int (8.6.10/8.6.10) id OAA27645 for ; Thu, 24 Jul 1997 14:27:26 -0400 Received: from domp04.adm.intelsat.int(164.86.130.139) by intelsat1 via smap (V1.3mjr) id sma004874; Thu Jul 24 13:26:15 1997 Received: from localhost (kimh@localhost) by domp04.adm.intelsat.int (8.6.9/8.6.9) with SMTP id SAA11771 for ; Thu, 24 Jul 1997 18:28:29 GMT Date: Thu, 24 Jul 1997 18:28:28 +0000 (GMT) From: Hong Kim Reply-To: h.kim@intelsat.int To: freebsd-questions@freebsd.org Subject: Re: Use of CVS for CMU CL (fwd) Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-questions@freebsd.org X-Loop: FreeBSD.org Precedence: bulk At a suggestion from Martin (from the CMU CL) project. I am writing about some advice on use of CVS. My message and his reply are below. I guess my questions center on 1) A reasonable branching strategy. Martin claims that the experience of the BSD project leads one to do development on the main branch (HEAD) and use branches to do I&T (Integration and Test) and eventually deployment. 2) Remote development. Client/server, trade patches, cvs import? What is good? What works well? What are the caveats to get it working good? Hong ---------- Forwarded message ---------- Date: Thu, 24 Jul 1997 17:38:27 +0200 (MET DST) From: Martin Cracauer To: h.kim@intelsat.int Subject: Re: Use of CVS for CMU CL Hello, > For example, on branching. I see two basic models. One uses the main > branch (HEAD) for blessed releases -- implying that new development and > transition through I&T will be done on branches. The other says that the > main branch is for active development and the I&T/deploy cycle are done on > branches. I don't know what I&T means, but I assume, a stable release. In CMUCL, we use the HEAD branch for the experimental stuff and brnaches for stable. Our model follows what FreeBSD does and those experiences show clear disadvantages when activly developing on anything else than HEAD, that means that the non-HEAD brnaches should get mostly merged-in diffs from the HEAD branch, but not too much new material. > Also, we are just starting to face the situation where development will be > done at remote sites and I would like to know what the CMUCL crowd has > done to reincorporate these remote changes. Client/server, patches, tar > files, ???? We use client/server over ssh with no problems. Of course, that requires that the outside groups a) are willing and capable of using CVS, too and b) that you trust them enough to let them modify your main CVS tree. If not, importing things from outside works good ("cvs import"). > It will also be additional ammunition to have shown a happy customer to > convince the Doubting Thomas' about the transition to CVS and the uses of > branching and tagging. Tagging, sure, but branching in CVS does not give what one might expect. You get close branches of one tree and you can exchansges fixes between those, but one of them is clearly the one for new development. The other's are not as capable. But remember that I don't run into these problems myself, I just avoid them based on the experiences the FreeBSD folks made. You might want to ask those about it (on freebsd-questions@freebsd.org or such) if you want to be certain. But then, what are the alternatives? Martin -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Martin Cracauer http://cracauer.cons.org Fax +49 40 522 85 36