Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 23 Oct 2015 13:53:47 +0300
From:      Gleb Smirnoff <glebius@FreeBSD.org>
To:        Bryan Drewery <bdrewery@FreeBSD.org>
Cc:        Hans Petter Selasky <hps@selasky.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r289700 - in head/sys/compat/linuxkpi: . common common/include common/include/asm common/include/linux common/src
Message-ID:  <20151023105347.GQ97830@FreeBSD.org>
In-Reply-To: <5627B24A.4010905@FreeBSD.org>
References:  <201510211501.t9LF1phM000793@repo.freebsd.org> <5627AF01.8070501@FreeBSD.org> <5627B1B5.30607@selasky.org> <5627B24A.4010905@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Oct 21, 2015 at 08:42:02AM -0700, Bryan Drewery wrote:
B> >>> Log:
B> >>>    Start process of moving the LinuxKPI into the default kernel build by
B> >>>    creating an empty directory tree.
B> >>>
B> >>>    Sponsored by:    Mellanox Technologies
B> >>>
B> >>> Added:
B> >>>    head/sys/compat/linuxkpi/
B> >>>    head/sys/compat/linuxkpi/common/
B> >>>    head/sys/compat/linuxkpi/common/include/
B> >>>    head/sys/compat/linuxkpi/common/include/asm/
B> >>>    head/sys/compat/linuxkpi/common/include/linux/
B> >>>    head/sys/compat/linuxkpi/common/src/
B> >>>
B> >>
B> >> This makes me think a branch should be used instead.
B> >>
B> > 
B> > Most of what will be done is "svn mv" out of "sys/ofed" and changing
B> > some include paths in sys/modules and sys/conf . Can you explain why you
B> > think a branch is required ?
B> 
B> Because you're splitting commits up that should otherwise not be split
B> up. There's no reason to commit empty directories before a 'svn mv' into
B> them.

+1

Any commit to head should be an atomic change that does something useful:
fixes smth, adds something useful, or removes something obsoleted.

You can either bring in the linuxkpi to head in a single commit, or, if
you prefer to do a serie of commits, where intermediate state is rather
meaningless, then you should do this serie in a separate branch, and then
produce patch from the branch and commit it to head.

There is no need to revert r289700. This is just a note for future.

-- 
Totus tuus, Glebius.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20151023105347.GQ97830>