Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751077AbVI2DYu (ORCPT ); Wed, 28 Sep 2005 23:24:50 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751028AbVI2DYu (ORCPT ); Wed, 28 Sep 2005 23:24:50 -0400 Received: from mail2.asahi-net.or.jp ([202.224.39.198]:65094 "EHLO mail.asahi-net.or.jp") by vger.kernel.org with ESMTP id S1750808AbVI2DYt (ORCPT ); Wed, 28 Sep 2005 23:24:49 -0400 Message-ID: <433B5E2B.4030805@thinrope.net> Date: Thu, 29 Sep 2005 12:23:23 +0900 From: Kalin KOZHUHAROV User-Agent: Mozilla Thunderbird 1.0.6 (X11/20050913) X-Accept-Language: en-us, en MIME-Version: 1.0 Newsgroups: gmane.linux.kernel To: Jesper Juhl Cc: Pavel Machek , Andrew Morton , linux-kernel , Alexey Dobriyan Subject: Re: [PATCH] Documentation: getting and installing git References: <200509290305.01625.jesper.juhl@gmail.com> <20050929020221.GN30883@pasky.or.cz> <9a874849050928193437a176bc@mail.gmail.com> In-Reply-To: <9a874849050928193437a176bc@mail.gmail.com> X-Enigmail-Version: 0.92.0.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 6473 Lines: 188 Just my 2 yen below: Jesper Juhl wrote: > On 9/29/05, Petr Baudis wrote: > >> Hello, >> >> I think Pavel Machek had some short git HOWTO as well, so I'm Cc'ing him. >> > > Ok, thanks. I was not aware of that. > > > >> In general, I think it's not a good idea to just duplicate GIT >>documentation in the kernel tree. If you think the GIT documentation is >>insufficient or is missing some "quick start" document, it'd be more >>reasonable to submit patches for GIT, but I'd keep kernel's GIT >>documentation to kernel-specific usage and tips'n'tricks. >> > > The reason for this document is to cater to the people who have a > kernel source tree available, know the kernel is managed with git, but > don't know how to obtain it. Such people (myself included) probably > take a quick look in linux-/Documentation/ for a document > telling them how to get and install git. This document is written to > serve those people. > > > >>Dear diary, on Thu, Sep 29, 2005 at 03:05:01AM CEST, I got a letter >>where Jesper Juhl told me that... >> >>>--- /dev/null 2005-09-28 20:05:57.000000000 +0200 >>>+++ linux-2.6.14-rc2-git3/Documentation/get-and-install-git.txt 2005-09-29 02:57:59.000000000 +0200 >>>@@ -0,0 +1,130 @@ >>>+ Getting and installing `git` and pulling your first tree >>>+ -------------------------------------------------------- >>>+ >>>+ (Writen by Jesper Juhl, September 2005) >>>+ >>>+ >>>+This document describes how to obtain and install the `git` tool used (among >>>+other things) to manage the Linux kernel source tree. It also shows you how >>>+to use git to pull down your first copy of the vanilla Linux kernel source >>>+(current git head version). >> >> Since you're cc'ing me, you'll get a shameless plug. ;-) What about > > > I'm Cc'ing you exactely because I thought you would have some valuable input :) > > > >>some decent short notice like: >> >> Note that you might prefer to use one of the simpler user interfaces >> available for GIT, e.g. the Cogito layer or StGIT patch manager. See >> the GIT homepage for details. >> > > something like that would probably make sense, yes. > > > >>>+Those who already have an older version of git can grab a newer version with >>>+ it clone http://www.kernel.org/pub/scm/git/git.git LOCALDIR >> >> Missing leading 'g'. >> > > Whoops. > > > >>>+To obtain the latest git source snapshot go to this URL: >>>+ http://www.codemonkey.org.uk/projects/git-snapshots/git/ >>>+and download the latest version (at the time of this writing the exact filename >>>+is git-2005-09-29.tar.gz, but a symlink called git-latest.tar.gz is also >>>+provided that will always pull the latest git source regardless of its actual >>>+filename). >> >> I think recommending the latest development snapshot instead of the >>latest release is a really bad idea if you don't have some really >>compelling reason to do so. The snapshot can be variously broken and >>buggy, while a release gives you some stable reference point and path >>from it you can follow. >> > > Hmm, yes, you are probably right. I should recommend the latest > release and just provide a small note on how to get the latest > snapshot. Thanks. > > > >>... >> >>>+At this point you should have git installed and available in your PATH. >> >> Perhaps you might rather want to extend GIT's INSTALL file? >> > > I can provide a patch for that if you like, sure. > > > >>>+Now it's time to download your first kernel source tree. To do that you should >>>+first change into the directory where you want to store the kernel source in a >>>+subdir. I'll assume you want to keep kernel source in ~/linux-kernel, so do >>>+this : >>>+ >>>+ $ mkdir ~/linux-kernel ; cd ~/linux-kernel >>>+ >>>+Now let's use git to download the latest git HEAD (the current head of Linus' >>>+development tree). Execute these commands to do this : >>>+ >>>+ $ git clone \ >>>+ rsync://rsync.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git \ >>>+ linux-2.6 >>>+ $ cd linux-2.6 >>>+ $ rsync -a --verbose --stats --progress \ >>>+ rsync://rsync.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git/ \ >>>+ .git/ >> >>Why the second rsync command? If you are after tags and other heads, you >>can run it just on .git/refs/. >> > > Hmm, right, my bad. I am not at all versed ing git, but AFAS rsync is considered, wouldn't it be times faster to start from a tarball (say latest -rc) and then rsync. Most people have it on their machines or it is cached all around the net. Even if they don't have it, getting a tarball to prime the rsync will be a good idea IMHO. >>But actually, it is very dangerous. Never ever run it later than right >>after the initial clone (ignore what the "Kernel Hackers' Guide to git" > > > This is supposed to tell people how to get and install git for *the > very first time*, so this would be the initial clone. > > > >>tells you!). If you did any local commits, it will likely trash them, >>and if you didn't, it will probably completely confuse the tools which >>care about updating your working tree with new changes. >> > > I think the best thing is just to drop that second bit. > > > >>>+When the download finishes you'll have a brand sparkling new git HEAD linux >>>+kernel source tree in ~/linux-kernel/linux-2.6 >> >>[Nitpick] I'm not a native English speaker, but I think "brand new >>sparkling" is more right. >> > > I'm not a native english speaker either, so you may very well be right > - I honestly don't know :) > > > >>>+If you want to do a git bisection search to find what patch caused a problem, >>>+please see the Documentation/git-bisect.txt document in the git source tree. >>>+You may also want to read and/or use Documentation/git-bisect-script.txt >> >>This notice would be quite useful in the rather antique >>Documentation/BUG-HUNTING file. >> > > Not a bad suggestion - anyone else have an oppinion on that? > > > Thank you for your constructive feedback. Much appreciated. Kalin. -- |[ ~~~~~~~~~~~~~~~~~~~~~~ ]| +-> http://ThinRope.net/ <-+ |[ ______________________ ]| - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/