Re: iptables release

[Jan Engelhardt <jengelh@computergmbh.de>]

On Dec 1 2007 02:37, Yasuyuki KOZAKAI wrote:
>
>$ svn log -r5092 -v http://svn.netfilter.org/netfilter
>------------------------------------------------------------------------
>r5092 | (no author) | 2004-10-11 22:40:52 +0900 (Mon, 11 Oct 2004) | 1 line
>
>This is an empty revision for padding.

While empty commits can happen (it does take a specific course of action
though), the log message is unusual, esp. with (no author) and "This is an
empty revision for padding".

>------------------------------------------------------------------------
>
>** What I did to create iptables git tree
>
>I did
>	$ git-svnimport -v -s 702 -l 3070 -T iptables/trunk	\
>		http://svn.netfilter.org/netfilter 
>
>I thought that such argument to -T was not expected by the author of
>git-svnimport, but it seemd to work. ;) I did not use -b and -t for
>branches and tags. Because I found that some commits on 'origin' were
>broken by them.
>
>Sometimes the TCP connection to SVN server was disconnected while importing.
>In that case, I reset 'origin' and 'master' branch so that they refered
>the last suceeded commit, removed .git/SVN2GIT_HEAD and .git/ORIG_HEAD,
>and excecuted git-svnimport without '-s' again.

I would have rsynced the svn tree off, then do a local svn-to-git transform.
That would have avoided potentially dangerous disconnects.

>** A Comment
>Well, I don't want to do this again :)

Heh.