From owner-freebsd-ports Fri Jan 12 9:52:20 2001 Delivered-To: freebsd-ports@freebsd.org Received: from blues.jpj.net (blues.jpj.net [204.97.17.146]) by hub.freebsd.org (Postfix) with ESMTP id 3FC5537B400; Fri, 12 Jan 2001 09:52:02 -0800 (PST) Received: from localhost (trevor@localhost) by blues.jpj.net (8.11.1/8.10.0) with ESMTP id f0CHq1727689; Fri, 12 Jan 2001 12:52:01 -0500 (EST) Date: Fri, 12 Jan 2001 12:52:01 -0500 (EST) From: Trevor Johnson To: Satoshi - Ports Wraith - Asami Cc: Will Andrews , , Subject: Re: overzealous cleaning of Attics in ports tree In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-ports@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > This is customary in the ports tree. As you may have seen in the > past, the ports tree will undergo a periodic Attic cleansing. I never noticed it before. I suppose I haven't been keeping a copy of the repository long enough, or I didn't watch cvsup carefully enough. I don't see it mentioned in the Porter's Handbook, nor the Committer's Guide. > There will be way too many dead files lying about otherwise. It costs me five cents to store the entire FreeBSD repo on a CD-R. Over time, the cost of disks and tapes has (doubtless with some exceptions) decreased. I have the impression that most people do not keep a full copy of the repository, only the checked-out sources or a release CD-ROM. Supposing that's true, only a few people store the dead files. > (By the way, the reason why I am so adamant about having repo-copies > done when a replacement port enters a tree is precisely because of > this.) Sometimes files can be sent to an Attic, yet no replacement is made. Such files can still have value. For instance, the AUIS port was sent to the Attic in October of 1999 because it would only compile in a.out format. IMO the files could still be useful. For instance, someone might want to compile it under FreeBSD 2.x and make the binaries available. I know that porters aren't officially supposed to support 2.x (sometimes I wish that applied to 3.x as well ;) ) but the minimal support of keeping old files around--so someone can check out the ports tree as it was in the past--is of value IMO. I know that with 4.x, "make release" requires some ported software: docbook and jade come to mind. Suppose that one of those ports was killed, or even just renamed. If the policy of purging dead files from the ports tree remains in place, people would no longer be able to build old releases of FreeBSD. Being able to do so has value. There was a recent discussion on freebsd-current ("Fixing a.out compatibility") in which the need emerged for someone with a 2.x box to build ld.so on it. Anyway, as for AUIS again, someone might want to make a port of just the fonts from it. Someone might want to make a port of AUIS and notice that (after this purge) none exists. Having the dead files would give such a person a head start, or at least alert him to the ELF problems. If you look at the AUIS files in the tarball I made, I think you'll agree that this was not a trivial port. Another example is the gyve ports. They were killed because Ade Lovett wanted to get rid of GTK+ 1.0. The GYVE ports had not been updated in a long time, so they still used that old library. It looks to me as though the reason they had not been updated is that no one had noticed the project's new site, www.gyve.org. I've not looked at it closely, but I have a hunch that the current GYVE will work with a more recent GTK+. There's a probability that the dead files would be of use for someone who wishes to make an updated port. As the maintainer of aumix (http://jpj.net/~trevor/aumix.html), sometimes I look at old versions. Compiling them on FreeBSD is easier with the port. Some of them required patches, all of which have been purged. If I allow the purged files to vanish from my own copy of the repository, it will become slightly more difficult for me to support FreeBSD. Another example from the tarball I made is the rzsz port. I would liken it to a sign on a repository for radioactive waste. It should be kept around for a long time, to warn people of the danger. > That said, it seems like indeed Peter got a little overzealous. I > thought we were supposed to only delete files that have been in the > Attic for some time, not all Attics. I think my examples show that the length of time since a file went to an Attic is not necessarily a valid metric. I ask you to change the policy and restore the files which have no replacement. Again, I am willing to help. > Doesn't this cause problems for people who checked out the tree > recently (and since have a file moved to the Attic)? I've seen a message or two from people who were surprised by all the deletions. At least one seemed nervous about it. There should have been a heads up. -- Trevor Johnson http://jpj.net/~trevor/gpgkey.txt To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ports" in the body of the message