http://git.samba.org/?p=vl/samba.git; a=shortlog; h=refs/heads/struct_stat

[simo]

On Fri, 2009-05-15 at 11:02 +0200, Karolin Seeger wrote:
> Hi Volker,
> hi Tim,
> 
> On Thu, May 14, 2009 at 07:28:13PM +0200, Volker Lendecke wrote:
> > Well, I don't really have a good technical argument for
> > getting it into 3.4. My only point is that I will definitely
> > have to maintain a 3.4 with that patch for at least a few
> > years. I can certainly do that, but I would prefer to not
> > having a patch of that size in a custom branch :-)
> 
> I can't rate whether this patch is very risky or not. If it's not risky at
> all, I would agree to put it into v3-4-test as it changes many lines and
> backporting might become difficult at some point. Volker, could you please
> rate if it's risky or not?
> 
> Of course Tim is right, v3-4-test is already feature freezed, but in this
> case an exception might make sense supposing the patch is very well tested
> and the risk is very, very low.
> 
> What do you think?

Unless someone is willing to test all vfs modules on all platforms I
think it is a bit risky. It's easy to miss a change and have modules not
compile (or worse fail at runtime).

As Tim pointed out, right now the stat struct is machine dependent, so
just having it build and properly work on one linux box does not
guarantee that it will build/work successfully on other architectures.

If we still have time to do reasonable testing it would probably make
sense, if we build master on the buildfarm and see no major
problem/regression in tests we might consider it safe enough.

But this is an exception to the rules we just recently stated so it
needs to be carefully considered. Out of guts feelings I'd say master
only, but I do not want to cause unnecessary work for other people
either ...

Simo.

-- 
Simo Sorce
Samba Team GPL Compliance Officer <simo at samba.org>
Principal Software Engineer at Red Hat, Inc. <simo at redhat.com>