Re: [RFD] Explicitly documenting patch submission

[Len Brown <len.brown@intel.com>]

On Sun, 2004-05-23 at 11:38, Ian Stirling wrote:

> Has anyone ever tried to forge the name on a patch, and get it
> included?

Yes.

Today akpm send me a little patch via e-mail, I did this:

$ bk import -temail < akpm.email

This records the author as akpm -- not me.
I did a "bk comments" to clean up the comments,
but the author remains akpm, who included a
single "Signed-off-by: Andrew Morton <akpm@osdl.org>"

If you paw through the s-file, you'll find a little [lenb]
that shows I checked in the file -- but the tools don't
seem to show that.  You'll also see a little [torvalds]
on lots of akpm csets, so I guess Linus does the
same thing.

There is a clear audit trail that the csets came from
my repo: http://lkml.org/lkml/2004/5/3/77
Though I guess pawing through LKML archives to follow
patch origin should be an exception rather than a rule.

More often I apply a patch with "bk import -tpatch".
Here I get recorded as the one who checked-in,
so if the author was not me, I credit the author
in the check-in comments.  But I suppose that
here too I'm a forger b/c I use BK_USER=len.brown
so that the history records my valid company
e-mail address, rather than the userid [lenb]
I've got on my local development box.

I guess this second method is consistent with Linus'
proposal -- though I would have expected the first
method to be preferable -- at least for well-known
authors.

Also, I guess the news here is that sometimes the
last two levels of the check-in chain are automatically
recorded, but this is not well known.

cheers,
-Len