This week in Bayonne and Common C++

[David Sugar]

Actually I have been very busy in two specific areas; one of them is the
upcoming GNU Common C++ 1.9.1 point release, which is supposed to fix
everything that was broken in 1.9.0 :).  The other is a 1.8.1 point
release of GNU ccScript.

In 1.9.1, we are trying to get Solaris targets working again, and I have
reorganized the headers and inlines so that we no longer expose so many
pthread functions as inline calls in the GNU Common C++ headers.  This has
always been a problem because various libc (including glibc)
implimentations have strange defines and requirements to compile these
correctly and consistently, and I would rather bury this in the compiled
sources rather than inline methods as we did before.

Also, more work has been done in making the 1.9.1 release usable native
under windows.  This is an emmense task in some ways because of the (in my
opinion) extremely stupid way windows expects dll symbol references that
will be exported to be defined and tagged in the headers when building the
dll, and defined and tagged in the headers when linking the other side, as
well as their attribute extensions to otherwise clean C++ syntax.  To deal
with this, we now make extensive use of specialized macros which Freddy
came up with that properly insert the dll import and export attributes
when compiling for a native win32 target.  In fact, Freddy is doing much
of the win32 target cleanup to make sure that these macros are now
properly used everywhere, and he has also started a Solaris test compile.

In GNU ccScript, I have been busy externalizing some functionality into
seperate DSO packages.  There will be both a "time" and "date"
manipulation package (.pkg) available as seperate loadable modules in the
next ccScript release.  The current limited built-in date and time support
in GNU ccScript will be removed.  Other core features may eventually be
re-introduced as external loadable modules.

Other changes in GNU ccScript are more subtle.  The syntax for labels has
been changed to support local labels in the form xxx: as well as ::xxx.
I also have a scheme for embedding pointer objects in ccScript symbols.
This will be used for special effects in Bayonne, such as for referencing
the object that holds the current mailbox in the new Bayonne messaging
DSO.

For GNU Bayonne, I will have to update the translators for speaking dates
and time based on the new ccScript date and time formats as created by the
new ccScript packages.  In addition, I am going to add the brooktrout
driver tree.  We now have a brooktrout card to play with, and I have a
complete set of their current api.

Mark has been able to get GNU Bayonne to run with the Dialogic CSP runtime
and an E1 card after what I gather has been a lot of work.  This will be
a very neat addition for the next release.  At least we now know it can be
made to work...

I have spoken with David Rowe about consolidating VPB support around the
standard Voicetronix API (we use a few special functions he added just for
Bayonne).  I do not know when this work will be started, but it should
have no effect on external use.  I also added some experimental code to
impliment Voicetronix call progress detection, and I have reorganized
support for CID, so that both occur more cleanly in Bayonne managed
sessions.

I see some interesting work in the LTAPI lists, and there is now a new 4
port analog dsp telephony card coming into the market based on the Zaptel
interface.  We need to track these changes in current GNU Bayonne driver
trees.

The next point release of GNU Bayonne will probably reflect current GNU
ccScript changes, perhaps some of Mark's patches for the Dialogic driver
tree, and perhaps also include the initial brooktrout tree, and the start
of the voice messaging and debit dso modules.

One thing I need now is people to test compiling GNU ccRTP under win32 to
see if it builds cross platform correctly with the current cvs edition of
GNU Common C++.  Another area of concern is having more people available
to work on digital development in the osdl facilitity.  Mark's work has
actually been done at his own facility with a live circuit.  Finally, I
need someone to help with maintaining the Bayonne web site.

David