[gdome] Fwd: Gdome
- From: Daniel Veillard <Daniel Veillard w3 org>
- To: gdome rpmfind net
- Subject: [gdome] Fwd: Gdome
- Date: Wed, 7 Apr 1999 19:21:08 -0400
Forwarded for achival,
Daniel
----- Forwarded message from Raph Levien <raph acm org> -----
> Date: Wed, 7 Apr 1999 13:26:44 -0700
> From: Raph Levien <raph acm org>
> To: Daniel Veillard w3 org
> Subject: Gdome
>
> Hi Ian and Daniel,
>
> Daniel sent me an email with a number of proposals. I agree:
>
> 1. *exc should be the last argument even on void-returning functions.
> Being consistent is more important than the small optimization of
> utilizing the return value.
>
> 2. Callees should set *exc = 0 on no-exception cases. This can be done
> in the invokers in gdome.c, so there's no extra code that needs to
> get written or bugs that can happen if it doesn't get done.
>
> 3. We should have a gdome mailing list. Daniel has volunteered to set
> it up.
>
> I have done an extremely rough sketch of a Corba interface in
> www.levien.com/free/rxml.tar.gz . I have deliberately not checked it
> into CVS because it's a mess - arbitrarily chosen names, global
> variables, etc.
>
> The utf-8 to utf-16 recoder is ok, but doesn't check enough errors. It
> should generate exceptions in this case :)
>
> The CORBA object creation stuff is ok. It shows the essentials of
> allocating space for the servant, initializing it, activating it,
> obtaining the CORBA Object reference, and putting it in a hash table
> so you don't get more than one CORBA object for each native object.
>
> It does not, however, do anything special about inheritance. I think
> the Right Thing to do is to make the most specific subclass possible,
> i.e. even if you return a Node, if the native object is actually a
> GdomeElement, a DOM_Element is what should be created. It looks like
> getting the type attribute of nodes is what we want to do here.
>
> The example does show the use of implementation inheritance, i.e. the
> vepv's for Document and Node both contain the node epv. This is as
> opposed to a "orbit-idl --skeleton_impl", which generates separate
> DOM_Element_DOM_Node epv's for each subclass of Node (and thus a >
> 6kloc C file!).
>
> I haven't played at all with the reverse direction, but it ought to be
> pretty straightforward - it would have a very similar structure as
> gdome-xml, but would use CORBA_Object pointers instead of xmlNode etc.
>
> Daniel also suggested I look at the gnome-dom stub generator, but I
> can't find it.
>
> Raph
----- End forwarded message -----
--
[Yes, I have moved back to France !]
Daniel Veillard w3 org | W3C, INRIA Rhone-Alpes | Today's Bookmarks :
Tel : +33 476 615 257 | 655, avenue de l'Europe | Linux, WWW, rpmfind,
Fax : +33 476 615 207 | 38330 Montbonnot FRANCE | rpm2html, XML,
http://www.w3.org/People/W3Cpeople.html#Veillard | badminton, and Kaffe.
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]