[asterisk-dev] Zaptel project being renamed to DAHDI

[Oron Peled]

OK, I'll chime in...

On Wednesday, 21 בMay 2008, Julien Goodwin wrote:
> On 20/05/08 21:53, Kevin P. Fleming wrote:
> > Actually, that's not the primary reason that Zaptel has not been pushed
> > for including in the kernel tree. The reason is that we still want
> > dual-license control over the code in Zaptel/DAHDI, and we can't easily
> > maintain that control if the code is managed in a tree we don't own.
> 
> *blink*
> 
> You don't *have* dual-license control over Zaptel. It's a kernel module
> *developed* for linux.

First, let's give credit where credit is due. Zaptel was developed for BSD
and later ported to Linux:
  http://www.zapatatelephony.org/

> One of the things the kernel devs generally agree 
> upon is that regardless of the situation for proprietary modules in
> general, those developed *for* linux (as opposed to ported to) *must* be
> GPL.

I think you extrapolated Kevin's talk about "control" into
a "we're gonna ship binary zaptel/dahdi..." which he didn't say.

While zaptel is GPL'ed, in practical terms most of its development
is done by a small group of people. Therefore, Digium maintain *some*
(informal) level of control (interface changes etc.) which they are
afraid to loose.

For the record, I personally don't buy the argument that pushing it
into the kernel would cause this group to immediately loose this type
of control. At least initially, it would be the same people and same
companies that would develop it.

I'll use this opportunity to touch a parallel subject of kernel
merging, namely, the "release-management-would-make-it-hell" problem:
  IMO, the panic of the release cycle mismatches between
  the kernel, zaptel and Asterisk is overinflated.

  There many subsystems within the kernel that went through a
  far more drastic changes -- take Wifi for example when the
  changes were/are very radical and involve vendors with totally
  conflicting interests. OTH, Zaptel had very minor interface changes
  both to user-space and to low-level drivers.

  While mainstream merging would initially cause some release
  hiccups, non of them would be tragic:
  1. As long as they are built as modules by the distros (pretty
     obvious). Someone that must be on the bleeding edge may
     replace them.

     [I can attest this is done daily as some of our clients need
      to regularly replace old drivers/firmware supplied by
      trixbox and other Asterisk oriented "distributions"]

  2. The kernel release cycle is very fast. Distros that track
     up to date kernels (Fedora, Ubuntu) are bound to satisfy many
     users (the rest may resort to 1.)

  3. Digium may setup an "official" git repository that would provide
     release candidates upstream (like most kernel subsystem maintainers
     do). Each company will have its bleeding edge tree for the latest
     and hotest hardware (which is the same as today, btw).

Merging into the mainline would provide enormous benefits
to the zaptel community. Mainly, it would force us to dramatically
improve its (pretty crappy) code quality (yes, I know a big part of
that is my fault...)

It's obvious why Digium isn't enthusiastic about it, but I think
they are making a mistake in the medium/long range, as they would be
a big beneficiary if every Linux out there would have zaptel by default.
Just think about the implications that any hardware released a
year ago would "just-work" on any current Linux distribution.

Bye,

-- 
Oron Peled                             Voice/Fax: +972-4-8228492
oron at actcom.co.il                  http://www.actcom.co.il/~oron
ICQ UIN: 16527398

"The speed of light really is too slow nowdays." -- Alan Cox