Hi!

> > He just thinks he can fix his code, and I want that code to be
> > reverted, reviewed, tested, and than merged back. There's no way
> > current mess can be fixed in reasonable time.
> 
> I realize that posting it before merging it would have resulted in a
> better outcome, as least thus far. However, it's been 11 days since I've
> merged the orgininal code, and 2 since I posted the last patch. While
> you've been ranting and raving, you've squandered a lot of valuable time
> that could have been spent reviewing the code. 

I've been reviewing your code, see your mailbox. Unfortunately due to
you renaming functions and moving files around, it is very hard to review.

> > I've seen no discussion about that code, and certainly have not seen
> > that code before it was merged, which is strange since I'm listed as
> > software suspend maintainer.
> 
> BFD. swsusp is one piece of the puzzle, and I've not touched any of the 
> core functionality. What I have done is: 

[Attached is patch "not changing core functionality". How did you
expect me to verify that? And it was you who protested on killing 3
printks.]

> Secondly, look at what I've done to it: 
> 
> - Compartmentalized portions that are useful to more general 
>   suspend/resume sequences. 
> 
> - Removed duplicate code between different PM mechanisms.
> 
> - Made it possible to use it with ACPI, instead of in lieu of it. 
> 
> - Cleaned up several functions, streamlined a few, added comments, and 
>   made the names easier to swallow than the 'magic' variety. 
> 
> - Removed 5 calls to panic() and 8 instances of BUG(). 

And managed to call sleeping functions with interrupts disabled and
break x86-64 somewhere in the process. Hmm, and because you killed
BUG_ON(in_atomic()), you did not realize that you were breaking
that. And I do not think you actually tested those "panic" codepaths
to make sure you are not corrupting data, right?

> In all actuality, I don't need swsusp. I have a better suspend-to-disk
> implementation that is faster, smaller, and cleaner. I've hesitated
> merging it because I thought swsusp improvements would be more welcome.
> Obviously they're not; or you haven't actually taken the time to read the
> code.

Swsusp improvements *would* be welcome, but I prefer to get them via
email, not via ftp from Linus.

> I will also restore swsusp to whatever state you like - either -test1,
> -test3 or -test4 state, or keep it the way it currently is in my patches.  
> But note that doing so will result in a large amount of duplicated code
> which you will be responsible for either merging or removing.

Good, please return it to -test3 state. If you can leave your split-up
patches on some public ftp site, that would be good; when dm is back
working so I can actually test it, I'll do some cherry-picking.

> Also, continuing to flame me without a rebuttal is not very polite, and
> indicates that you don't want to have an actual conversation. Please
> remove me and any public mailing lists from the cc list if you're going to
> continue to do this. I'll eventually filter you out, but please save
> everyone else from the drivel..

...while this surely contributes to actual conversation.
									Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]