From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from elasmtp-dupuy.atl.sa.earthlink.net ([209.86.89.62]) by pp; Tue Dec 2 02:34:07 EST 2014 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=ix.netcom.com; b=hu7eBnUEWbNE+JVNEAObneCodG346LsILazsIXSVJhUhKUSPpR6rdODTQ9hA3828; h=Received:Message-ID:Date:From:User-Agent:MIME-Version:To:Subject:References:In-Reply-To:Content-Type:Content-Transfer-Encoding:X-ELNK-Trace:X-Originating-IP; Received: from [65.60.177.124] (helo=[10.1.21.2]) by elasmtp-dupuy.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from ) id 1XvhyB-00081U-Th for 9front@9front.org; Tue, 02 Dec 2014 02:34:00 -0500 Message-ID: <547D6B66.7080300@ix.netcom.com> List-ID: <9front.9front.org> X-Glyph: ➈ X-Bullshit: scale-out TOR over TOR framework markup Date: Tue, 02 Dec 2014 02:33:58 -0500 From: Marty User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Icedove/31.2.0 MIME-Version: 1.0 To: 9front@9front.org Subject: Re: [9front] Debian port of Plan 9 References: <37cc8303be6963761b9e34c01901d0db@xx230.inri> In-Reply-To: <37cc8303be6963761b9e34c01901d0db@xx230.inri> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit X-ELNK-Trace: d195420e111396b06f36dc87813833b225aa8a2065c9591f5151a62e849e485669ee95add54a9959350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 65.60.177.124 On 11/30/2014 11:09 PM, sl@9front.org wrote: > Welcome to the 9front mailing list.[0] > >> I'm still new to Plan 9. My current desktop is Debian, so I am looking >> for way to combine both interests. Before I go on, I see the Debian port idea opened a can of worms. It was just an idea, and not a well considered one. > Why? I've been wanting to try Plan 9 for a long time. I want to see what the Unix developers came up with, to replace Unix. Everything I read and hear about it sounds good. > Have you installed Plan 9? I'm very late to this party. I just downloaded several ISOs to try. What do you do with it? If it works out, I hope to set up several boxes, replace my nfs network, fill in any missing software as time permits, and learn about kernel development. That should keep me busy for awhile. What do you think is > missing? It looks like it could use more drivers. I might be also able to help with small kernel patches or arch ports. That would be nearly impossible in Linux these days. My background is mostly hardware design and embedded software. (No Windows or Mac.) I've also done some Linux and other *nix drivers. >> How feasible would it be to do a Debian port of Plan 9, like the Debian >> GNU/Hurd experimental port, and how suitable would 9front be as a >> starting point? > > What would be the purpose of doing a Debian port of Plan 9? Their archive, infrastructure, and expertise might be of use, and I think a lot of frustrated developers might be interested. I saw it as a bridge for people to move away from *nix to something better, without leaving Debian. Now I'm not so sure. > 9front would not be a good starting point for a Debian-style fork of > Plan 9, because 9front's licensing is essentially identical to that > of the Bell Labs release. Nobody has bothered to sort out a specific > licensing scheme for new code. (Mainly because the people writing the > code don't care about licenses.) This is probably not compatible with > your goals. It sounds like a red flag, but I don't know much about licenses. > You might want to check out Akaros[1], an an open source, GPL-licensed > operating system for manycore architectures that includes a GPLv2-licensed > copy of the Plan 9 (not 9front) source code. Thanks for the tip. >> A Debian port would entail: >> >> - multiple architecture support > > 9front currently runs on the following platforms: > > 386 > amd64 > arm That looks pretty good. Debian Hurd only has 32 bit x86. >> - porting the packaging system (dpkg/apt) > > Why? Compatibility (dependencies) mostly. I think Plan 9 features like union directories and namespaces might eliminate or change the requirement, and some Linux distros are experimenting with similar ideas. I am also looking at JIT compiling and compile-from-source. I suspect the distro- as-packager model is obsolete. Debian will have to change here and an experimental port seemed like the place to test it. >> - porting all the apps > > What apps? And why? Most of this stuff is not wanted by Plan 9 users, > and in any case is grossly incompatible with Plan 9's interface conventions. > If you need these programs and enjoy the way most of them isolate you from > the general features of the operating system, why not just use the operating > systems where they already live? I think you're right. Appealing to the lowest common denominator of users is problem in Debian right now (systemd issue in a nutshell). I would not want to see it in any new port. Popularity is a double-edged sword, but I don't know how to prevent it, or prevent it from corrupting development. >> - adding the missing drivers > > Do you understand that Plan 9 is not Linux? Do you understand that Plan 9 > is not even UNIX? New drivers need to be written from scratch. Yes, yes, and yes, drivers are a serious commitment of time and effort, but I want to see what it's like to work with Plan 9's kernel. That's the part the interests me the most. >> - following Debian policy and release schedule > > What is Debian policy and what is Debian's release schedule? Why should > Plan 9 users follow them? What benefit is delivered by all this bureaucracy? I think Debian policy is mostly about criteria for package acceptance: https://www.debian.org/doc/debian-policy/ I don't like the bureaucracy either, but I was hoping that an experimental port would get minimal interference. Could be wishful thinking. > Are you aware that the Bell Labs Plan 9 ISO image is already automatically > generated, every day?[2] Thanks, I didn't know that. I think the site was down when I checked. >> - finding Debian developers who will sponsor and sustain the port > > Are there any Debian developers who have ever used Plan 9? None that I know, and while looking, I just found this: https://lists.debian.org/debian-devel/2011/08/msg00619.html So it looks like idea that has already been proposed and rejected. Now seems like an even worse time to propose it. Why would > they want to? And why would Plan 9 users want to put them in charge of > Plan 9? It's a good question and I never looked at it that way. Debian was a small and vulnerable project when I started using it, but in reality it's a 600 pound gorilla. How a port might affect or even harm Plan 9, or 9front, is a serious question. I think 9front have answered it, and now I understand that. Projects should probably stay clear of the trainweck. >> I don't know anything about getting approval for such a project, but >> if it happens I think there will be a lot of interest. Debian seems like >> a natural fit because it is the only distro (that I know of) that >> supports alternate OSs and kernels, and Plan 9 would be the first to >> do both. The port could also be done unofficially, but it would have >> less visibility and support. > > I don't understand how bolting a Debian logo onto Plan 9 would improve > the situation we currently find ourselves in: > > A world drowning in terrible software, much of which is propagated by > various Debian-related projects. > > Adding more programmers and/or more users usually just results in adding > more garbage to the operating system. Most Plan 9 users are not interested > in that. I agree, I was only looking at it from the perspective of getting developers, not non-technical users. >> There are many porting issues. A large number of Debian apps use dbus, >> so that might be a biggest obstacle to porting (this is a general >> problem in Debian ports). Could 9p be used instead as a dbus >> replacement? I don't know much about either protocol, but 9p seems like >> a more professionally engineered solution. > > This last bit raises a flag. Why are you suggesting a Debian port of > Plan 9 if you don't already know very much about either dbus or 9p? Are > you one of those people who proposes a massive project (note: staffed and > funded by others) after cursory evaluation of things that seem "neat"? Just asking questions at this point. I am not even close to proposing anything, but I think there is a lot of interest. > It also rasises a flag that you seem to think porting Linux programs to > Plan 9 is a desirable goal, in general. You seem to be suggesting that > the benefit is so obvious that it doesn't need to be explained or justified. I see your point. I didn't give that statement much thought. >> Unix APIs, POSIX and FHS are also porting issues to contend with, but I >> am hoping the Plan 9 for User Space will alleviate some of those. For >> graphics apps, could ports be done from one of the frame buffer >> ports, like the Debian ARM port? I have more questions, but I'll stop >> there. > > How would a project that combines ports of some Plan 9 utilities with > attempts to replicate Plan 9 functionality using native tools on POSIX > systems help alleviate API, POSIX and FHS issues associated with porting > UNIX programs to Plan 9? I totally misunderstood its purpose and I have no interest in porting Plan 9 apps to Linux. That should have been framed as a question. > It really sounds like you don't know every much about Plan 9 at all, > and maybe not as much about Linux and UNIX as you might think. This is > not a sin, but is also not grounds for proposing a Debian port of Plan 9. Sorry if I gave any other impression. I am a just starting out and asking questions. I'm also not a userspace guy, and I guess it shows. Plan 9 looks like userspace done right and that intrigues me. > You're probably going to disagree with some of the things I wrote above. > The best way to work through this disconnect is to use Plan 9 for yourself > and see how you enjoy the experience. It could even turn out that Plan 9 > is not for you. I will give it a shot. Thanks for your reply. This was very helpful. > In the meantime, to aid in your research, quite a lot of information about Plan 9 has been collected in the 9front FQA[3]. > > sl > > [0] http://9front.org/img/welcome.jpg > [1] http://akaros.cs.berkeley.edu/akaros-web/news.php > [2] http://plan9.bell-labs.com/plan9/download.html > [3] http://code.google.com/p/plan9front/wiki/fqa >