Re: Why write an Ada web browser ?, was: Re: GNAT Ada - DLL - MSVC

[David Marceau <davidmarceau@sympatico.ca>]

Preben Randhol wrote:
> 
> Jeffrey Carter <jrcarter@acm.org> wrote on 13/06/2002 (00:41) :
> >
> > I would like a browser/mail reader/news reader that is more secure than
> > the existing ones. Engineering it in Ada would be very helpful in
> > achieving that.
> 
> Why on earth do you want to put all these things into single
> application? I would rather have two optimised applications that can
> talk to eachother then one single application. One reason is that each
> project would be smaller. The other is that when I use a web browser I
> don't want to be forced to also have its mail reader installed as I
> might want to use some other program or visa versa. At the moment I use
> Galeon (a web browser only) which uses the Mozilla redering engine.
> 
> --
> Preben Randhol         �For me, Ada95 puts back the joy in programming.�
I apologize for not being clear.  I meant different applications: 
ada-web-browser, ada-mail-reader, an ada-news-reader.
That said the underlying infrastructure. i.e. sockets, ssl,
widgets(gtkada like) it's understood could be shared among them.

Concerning your reference to Mozilla, IMHO this dream ada-web-browser is
not just a re-write of Mozilla in Ada; 
it's an overhaul of the architecture with different priorities in the
requirements.
Some algorithms could be inspired from Mozilla however 
the ada-web-browser architecture must ensure security, reliability and
long-term maintenance 
are the top-priority requirements for every service in every api.
IMHO as a consequence of in-process XP-COM being insecure, if any XP-COM
is used 
in the ada-web-browser-architecture at all, it should be out-of-process.
I've explained why out-of-process is better in a previous CLA email.

What are the first steps in order to encourage this further?
I would recommend having a good understanding of Mozilla core though
before embarking into this.
Deciding which gui with ada is a good question.
I would propose whatever is written should be isolated from the gui
anyways in order to switch gui sdks for different reasons.
I could name a few good reasons for this: 
1)embedded system single purpose gui target and 
2)Win32 and X-Window targets
Note how I intentionally left JVM out of this.  
The infrastructure built on top of target 1) and 2) will deal with
secure out-of-process plug-ins compiled for the target in-mind.
I like the ada mentality of dare I say "write once, compile everywhere
once" :)  But who's kidding who.  Reality is there are many versions of
JVM for every machine/OS(i.e. phones(j2me), smartcards(javacard),
pcs(java).  There are many versions of java/JNI for every OS.  
There are many versions of gnat for every OS.  There are many versions
of gcc for every OS.  There's a lot of compiling going on using any
language.

I would have peace of mind running an ada-compiled embedded system that 
could download out-of-process plugins and and run them as another
process with less privileges on the phone/handheld/javacard/pc....
Among the plug-ins could be an ada-web-browser.

But remember this is all a foggy dream :)

Sant� bonheur,
David Marceau