[Pipet Devel] Offer to help

Brad Chapman chapmanb at arches.uga.edu
Mon Oct 9 16:45:46 EDT 2000


Steve wrote:
>  I got into python last night, and I'm getting it pretty quick. Are
>  there any python-CORBA documents lying about? 

There aren't really any all inclusive docs/tutorials for learning CORBA
from python but here is how I learned to do it:

1. I didn't know anything about CORBA, so I read a C++ book to get a good
foundation for the basic ideas of how it worked. Henning and Vinoski is
by far the best -- here's a slashdot review of it:

http://slashdot.org/books/99/07/12/1236220.shtml

2. I then took a look at the official mapping, which is pretty readable
and a good reference for later:

http://cgi.omg.org/cgi-bin/doc?ptc/00-04-08

Be warned though, the only python ORB implementation that currently
follows this is omniORBpy (which is used in Piper, of course :-), so you
will find some deviations in other ORBs.

3. Then I started playing around with Fnorb (http://www.fnorb.org). This
is an almost all python ORB, and has a nice booklet that walks you
through a hello world example.

4. Finally, I started working with omniORBpy (
http://www.uk.research.att.com/omniORB/omniORBpy/) and playing around
with the examples. There is some documentation for omniORBpy, but last I
checked it wasn't very helpful from the "getting started" perspective.
However, playing with the examples really helped me get into this a lot.

>From there, I think you should be ready to go :-). At least for me, I
struggled a bit at first until I suddenly had a moment of clarity where I
started to understand what was going on with CORBA, and from there I have
had no major problems (especially with python, because it is sooooooo
nice for CORBA!).

I do not recommend trying to learn by reading the gnome documentation for
C programming with CORBA. The C mapping is pretty hard to follow,
especially at first, and since CORBA is very OO all the way, python is a
better place to start, IMHO.

Steve wrote:
>  > The last item on the DL todo (easier to write XML), looks like a
>  > good entry point into piper and python (the auto-"XSL" for now). 

Jean-Marc wrote:  
>  I'm not sure the "easier to write XML" is still pertinent with the
>  changes we have done about a month ago. Even if it still is, it might
be too
>  soon for that.
>  ....Or maybe I just didn't understand what Brad meant by that. Brad,
>  what do you think? 

Well, the items on the DL to do were generally structured by order of
priority and timeframe (at least how it seemed to me at the time :-). In
general, I'm not exactly positive if the "easier to write XML" is exactly
relevant right now, because that is something that was meant to get into
place once a working system was in place. Jean-Marc and I did write a
nice perl script to do the work on converting the old Overflow C++
declaration syntax into C++ automatically (including support for comments
:-), so I'm not sure if more work needs to be done here right now...
Sorry, not to be too discouraging on this! It is definately important,
but will probably come into play later down the line.

Looking again at the list, I think the most important thing is connecting
the DL with the BL, being worked on by Jarl. I got a garbled ICQ message
form Jarl yesterday which suggested he was having some luck in this area,
so IMO, this is probably the best area to work on, since you will see
some results right away (which is always nice). I have already written
the skeleton code in the DL for communicating with the BL, but this will
probably need some tweaks and error fixes once the BL is up and running
with a CORBA server. So, if I were you, I would go after this area first.
I can definately be more specific about things, if you want...

Brad









More information about the Pipet-Devel mailing list