[m-users.] users Digest, Vol 107, Issue 14

[Volker Wysk]

Am Montag, dem 16.10.2023 um 16:20 +0200 schrieb Michel Vanden Bossche:
> 
> > On 16 Oct 2023, at 16:07, Volker Wysk <post at volker-wysk.de> wrote:
> > 
> > Am Montag, dem 16.10.2023 um 15:39 +0200 schrieb Michel Vanden Bossche:
> > > > On Sat, 14 Oct 2023, Matthew Delaney wrote:
> > > > 
> > > > > Hi, I'm new to Mercury but am interested in using it to replace
> > > > > Prolog
> > > > > in a project I'm just starting. This project will need a web-
> > > > > interface
> > > > > and backend DB connectivity. Reading through Mercury's
> > > > > documentation,
> > > > > it looks like the most straightforward approach would be to use
> > > > > CGI
> > > > > for the Web App part and the Foreign Language Interface for DB
> > > > > connectivity. Is that correct or are there other options
> > > > > available?
> > > > 
> > > > If you're using one of the Java or .NET web frameworks, you can
> > > > compile
> > > > your Mercury to a Java or C# library respectively and call it from
> > > > the
> > > > framework.
> > > > 
> > > > Julien.
> > > 
> > > Our experience with the Java backend is that it offers great
> > > performance.
> > > A Mercury application complied to Java is viewed from the outside as a
> > > standard Java application (Spring Boot…) and can be easily deployed in
> > > the
> > > cloud, using Kubernetes for scalability. Just in case…
> > 
> > I've programmed in Java many years ago. And I think that the asm_fast
> > grade
> > must be WAY faster than Java. I have two time critical Mercury projects
> > and
> > the performance is astonishing.
> 
> 
> Yes, of course, outstanding, but you would be surprised by the performance
> with Java, with the additional benefit of having the possibility to use
> all the Java ecosystem. In the paper attached (section 7 Benchmark), you
> see that the performance “penalty” for using Java is a factor 2.5, quite
> reasonable. This was done in 2010 and should be redone (NB - The link to
> the demo cloud.missioncriticalit.com/rule-demo is unfortunately not active
> anymore).

This really surprises me. But they don't state if they used the low-level-C
(asm_fast) or the high-level-C backend. The former is faster.

Happy hacking,
Volker