[asterisk-dev] dahdi to upstream?

[Shaun Ruffell]

On Thu, Dec 29, 2011 at 10:33:33PM +0200, Oron Peled wrote:
> On Thursday, 29 בDecember 2011 20:52:52 Russell Bryant wrote:
> > On Thu, Dec 29, 2011 at 12:42 PM, Oron Peled <oron.peled at xorcom.com> wrote:
> > > Currently there's no policy for deprecating support for old kernels
> > > in DAHDI-linux, so we have all kinds of code monstrosities to
> > > support anything from prehistoric kernels (2.6.8/9) to the latest
> > > and greatest.
> > > ...
> > This all sounds fine to me, but on a very related topic, when is it
> > time to start pushing DAHDI upstream?
> 
> I'm all for it, but please split this discussion off the current
> thread (that's why I changed the subject)

My unnofficial opinion: Based on what I know right now, I too
support pushing DAHDI upstream. However it's not my call and there
are still things to work out both technical and otherwise.

DAHDI Linux has been transitioning to a more kernel compatible form.
New code has been following the kernel coding conventions as opposed
to the Asterisk coding conventions and we've been using a git-based
workflow for a couple of years amongst the primary contributors with
idea that adopting a workflow more like the kernel would mean an
easier transition if DAHDI is ever in the kernel.

Before I would broach the subject again with those who have final
say I would like to see a few technical issues resolved:

1) DAHDI Linux and Tools needs to switch to git as the primary
interface for the entire community.

2) The third-party echocanceler APIs need quite a bit of work to
make them kernel compatible.

and 

3) All the remaining checkpatch.pl (kernel style) errors would need
to be cleaned up. Progress is being made here (without just simply
doing wholesale reformatting yet). You can see the difference just
in dahdi-base.c from 2.2.0 to 2.6.0 below.

checkpatch summary for dahdi-base.c in 2.2.0:
  total: 953 errors, 940 warnings, 8074 lines checked (23% of lines
  have errors or warnings)

checkpatch summary for dahdi-base.c in 2.6.0:
  total: 646 errors, 582 warnings, 10125 lines checked (12% of lines
  have errors or warnings)


I am not personally going to push for upstream inclusion until
those technical issues are resolved.

Additionaly, once the code issues are worked out there will be the
buisiness case to make and administrative issue to work out which
are not unlike those Jesse Brandeburg presented at a 2010 Linux
Symposium talk "Developing Out-of-Tree Drivers alongside In-Kernel
Drivers" [1].

[1] http://www.kernel.org/doc/ols/2010/ols2010-pages-35-40.pdf

Again, this is just my personal unofficial opinion,
Shaun

-- 
Shaun Ruffell
Digium, Inc. | Linux Kernel Developer
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
Check us out at: www.digium.com & www.asterisk.org