[LinuxBIOS] #33: Signed-off-by in trac is meaningless

Stefan Reinauer stepan at coresystems.de
Sun Nov 5 23:38:05 CET 2006


* Segher Boessenkool <segher at kernel.crashing.org> [061105 22:32]:
>  > Signed-off-by: and Acked-by: always corresponds to the latest  
> added patch, unless otherwise noted. Since each patch and each  
> comment has a timestamp, everything is clear, right?
> 
> So two patches by Uwe on a ticket with 3 comments -- yeah sure
> we can handle that.  Now how about a ticket with 12 proposed
> patches by 4 people, each fixing part of the problem, some superseded
> by newer patches; and all of that intermingled with some odd 100
> comments?  Can you still align that so that you clearly see or supposed
> "paper" track of origin of the code?

Not easily, because trac groups all attachments together instead of
packing them into their submission comments.

In such a case you should write the name of the patch you 
Sign-off or Ack. "Unless otherwise noted".

In theory, a patch that solves half the problem only should not get
Acked.

> Maybe this is all just a symptom of a greater problem: patches should
> always be presented with a proposed check-in comment, and the
> signed-off-by should be part of that comment.

Yes.

The check-in comment should be "all of the discussion done on this
check-in", so you get a non-ambiguous bidirectional connection between a
check-in and the discussion that led to this check-in.

If you go through the revisions of v1 and v2, you will find a lot of
stuff that nobody will ever be able to find out what it was good for.
This was fine when it was done, but it will not be fine for a high
quality "product". And with a userbase of 10mio we should aim at that.

-- 
coresystems GmbH • Brahmsstr. 16 • D-79104 Freiburg i. Br.
      Tel.: +49 761 7668825 • Fax: +49 761 7664613
Email: info at coresystems.dehttp://www.coresystems.de/




More information about the coreboot mailing list