Re: Please pull powerpc.git 'merge' branch

[Linus Torvalds <torvalds@osdl.org>]

On Wed, 22 Nov 2006, Paul Mackerras wrote:
>
>     [POWERPC] Revert "[POWERPC] Add powerpc get/set_rtc_time interface to new generic rtc class"
>     
>     This reverts commit 7a69af63e788a324d162201a0b23df41bcf158dd.
>     
>     As advised by David Brownell:
>     
>     http://marc.theaimsgroup.com/?l=linux-kernel&m=116387226902131&w=2

Btw, when doing a web-pointer like this, _please_ don't just leave it 
hanging like that.

For one thing, there's no guarantee that the web-server will continue to 
exist, or not move around the data. And even _if_ the data still exists 
and at that address, when looking at bugs of old history it's really 
irritating to have to fire up a web browser - not to mention that maybe 
you're on an airplane or other place without internet access. Git works 
fine disconnected.

So the rule is:

 - it's perfectly _fine_ to point to web addresses for "further 
   information" (especially things like bugzilla, of course)

 - but the commit message also has to stand on its own, even _without_ the 
   pointer. So at least add a sentence about why - even if the web pointer 
   would explain it more.

So even if you just say something trivial like

	As pointed out by David Brownell, that commit was just a total 
	thinko, and could never work.

at least it doesn't leave the commit message _totally_ without any 
context.

		Linus