xmlgraphics-fop-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Allan Erskine <a.ersk...@cs.ucl.ac.uk>
Subject Re: FOP with C++
Date Fri, 02 Mar 2001 01:42:45 GMT
Been debating responding too...I've been following the earlier thread, and
although a FOP outsider, I've rooted around a _little_ (trying to get
keep-with-next to work!), so I guess my main stance is as a C++ (games)
programmer who's just moved over to java.  So apologies if this seems like
hot air!

IMHO FOP could have a beautiful C++ implementation.  Its tree structuring
would favour C++ in some ways, assuming there was a decent mapping into
C++'s typing...and from what I've seen of the XSL specification, FOP could
be an application that might favour a *really* smart implementation.

I think a lot of C++ programmers would come out of the woodwork for a good
FOP design.  I think it would be satisfying, _very_ fast (whatever the
merits of Java, you can't argue with the byte-shuffling), and a great
exercise in the language.

(as well as the portability argument)

Surely FOPC can't start soon enough...a healthy synergy followed by healthy
competition....what could go wrong?

- Allan


----- Original Message -----
From: "Art Welch" <art_w@EASTPOINT.COM>
To: <fop-dev@xml.apache.org>
Sent: Thursday, March 01, 2001 7:18 PM
Subject: RE: FOP with C++


> I debated responding to this, because it does not really deal with FOP,
and
> it is the same old lame argument... But I will try to be brief.
>
> Java is not NECESSARILY slower than C++. This depends on the application
and
> the implementation. While it may be true that for some (perhaps many)
> applications it may be possible to eek out greater performance with a C++
> application than a Java application. With modern Java and JIT and/or
HotSpot
> compilation and proper design the difference may not be significant. For
> example the C++ implementation of Xalan was actually SLOWER than the Java
> version (I do not know if this is still the case). Unfortunately Java
earned
> a bad performance reputation early on when everything was interpreted,
JVMs
> performed poorly, and developers had not learned how to program in Java.
> These have been (or are being) addressed. With the compilation techniques
> and improved JVM implementations, there need not be a significant
> performance penalty for Java vs. C++. Of course programmers need to learn
> that Java needs some different techniques for better performing Java
> applications.
>
> Regarding a C++ implementation of FOP. I would expect that since it has
not
> been started yet (to my knowledge), by the time there was something
> runnable, the Java folks would be well along in enhancing the performance
of
> FOP-J. So in all likelihood, the Java version would actually perform
better
> than the C++ version. Sure the C++ version MAY catch up - but how long are
> you willing to wait? Even if it does eventually perform better, would the
> likely small difference be worth the wait?
>
> I said more than I was going to... sorry,
> Art (brevity is not my forte)
>
> -----Original Message-----
> From: Avula, Raj [mailto:RAvula@firstam.com]
> Sent: Thursday, March 01, 2001 2:00 PM
> To: 'fop-dev@xml.apache.org'
> Subject: RE: FOP with C++
>
>
>
> Hi Peter,
> Thanks for the information. I know that I can use JNI or CORBA(our
> system already using this).
>  But the reason why I don't prefer Java is because it is slower than C++.
> Our system generates reports
> in which time is the greatest priority of report generation process.
>
> Regards.
> Raj
>
> > -----Original Message-----
> > From: Peter S. Housel [SMTP:housel@acm.org]
> > Sent: Thursday, March 01, 2001 10:38 AM
> > To: fop-dev@xml.apache.org
> > Subject: Re: FOP with C++
> >
> > "Avula, Raj" <RAvula@firstam.com> wrote:
> > > Frank Neugebauer [SMTP:neuggs@hotmail.com] wrote:
> > > > FOP is not currently availble with C++. This list has discussed such
> > an
> > > > implementation, but I don't recall seeing a consensus.
> > > Thanks for the information. I am really looking forward
> > > to use FOP implementation in C++. Please let me know if
> > > any body started such implementation.
> >
> > It should be possible to use the JNI (Java Native Interface) from C++
and
> > link with the Java runtime to allow use of FOP from C++ code.  The
> > difficulty would probably depend on where you want the input to come
from.
> >
> > Cheers,
> > -Peter S. Housel-   housel@acm.org   http://members.home.com/housel/
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: fop-dev-unsubscribe@xml.apache.org
> > For additional commands, email: fop-dev-help@xml.apache.org
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: fop-dev-unsubscribe@xml.apache.org
> For additional commands, email: fop-dev-help@xml.apache.org
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: fop-dev-unsubscribe@xml.apache.org
> For additional commands, email: fop-dev-help@xml.apache.org
>


---------------------------------------------------------------------
To unsubscribe, e-mail: fop-dev-unsubscribe@xml.apache.org
For additional commands, email: fop-dev-help@xml.apache.org


Mime
View raw message