The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

App::faq - App-Context Frequently Asked Questions

INTRODUCTION

This is the FAQ for the App-Context software framework (a variant of the Perl 5 Enterprise Environment). You can find out more background to the project on the web.

  http://www.officevision.com/pub/p5ee
  http://p5ee.perl.org

GENERAL

Why should I use App-Context rather than J2EE or .NET?

Perhaps it's because you prefer writing in Perl? But besides that, there are other reasons.

Java's promise of "write once, run anywhere" is actually better fulfilled by Perl than Java.

The .NET CLR's promise of "any language, one runtime environment" is great. But Parrot will do for dynamically typed languages what the .NET CLR does for statically typed languages. (See http://www.parrotcode.org/faq/.) And Perl6 will support the CLR as well.

I looked around at the state of various cross-platform application runtime environments, and saw that the most pervasive execution environments, available across many platforms were: Perl (on servers), Java (on servers and browsers), and Javascript (on browsers). Each of these technologies holds the promise, to some varying degree, of "write once, run anywhere". ".Net" was not yet on the scene, but even when it arrived, the promise of proper cross-platform support for .Net was a long way off and altogether uncertain.

Perl has excellent support even for unprivileged accounts at ISP's, whereas Servlet support is hard to come by unless you own the machine (or you go to a very specialized ISP). Also, Perl offers several ways to do web applications: CGI for unprivileged, quick and dirty implementations, and mod_perl for high performance implementations when you have full control over the web server (also PerlEx, FastCGI, etc.).

If you have a desire to program in Java and you like the API's that Sun (and others) have created, you should probably focus on Java. I think that things could be a lot simpler (or maybe higher level) than the J2EE specifies them.

It seemed that the one thing that Perl was lacking was a blueprint for large-scale development and deployment of high-performance, high-availability systems (i.e. enterprise systems) along with guides of discipline for coding and documentation. The App-Context framework fills this gap.

So that's the explanation of "why App-Context?"

For an explanation of "why not App-Context?" you might consider that it is still largely vaporware.

How well does the App-Context fit into a .NET technical strategy?

It may seem that App-Context is most at home with the following technologies.

  * Perl, Linux, Apache, CGI, and MySQL.

However, the App-Context framework abstracts much of the runtime environment, so that it is just as easy to use:

  * Perl, Windows NT, IIS, ISAPI, and SQLServer.

So the question becomes, "How well does Perl fit into a .NET technical strategy?"

Although Perl is not a core .NET language from Microsoft's perspective, the following three considerations would suggest that Perl is not outside of a .NET technical strategy.

 1. Perl integration with Windows and .NET is substantial 
    (using the ActiveState port of Perl)
    http://www.activestate.com/Products/ActivePerl/

 2. Perl's predominant SOAP implementation is very mature
    http://www.soaplite.com/

 3. When Perl 6 is complete, both Perl 5 and Perl 6 will be able
    to run on the .NET CLR.
    http://www.parrotcode.org/faq/

What is the relation between App-Context and mod_perl?

The people who have been developing and working on mod_perl have done an incredible service to Enterprise Perl in general, and therefore the App-Context in particular. Other efforts that dramatically enhance Enterprise Perl are SOAP projects, XML projects, DBD/DBI::*, Templating projects, etc. These are all making dramatic contributions, and they need not change anything they are doing in order to continue to do so.

On the contrary, the responsibility is on the designers of the App-Context to accommodate and incorporate these many enterprise-class technologies into an integrates whole. Browse the classes envisioned for App-Context to see one way that these might fit together. Take special note of the "Classes (Planned)" at the bottom of the "All Classes" frame.

  http://www.officevision.com/pub/p5ee/software/htdocs/api/

As for mod_perl, it would seem to be the container of choice for web applications and SOAP services.

  http://www.officevision.com/pub/p5ee/software/htdocs/Appx/Blue/Context.html

There is no need for the mod_perl project to explicitly integrate with App-Context. However, in the spirit of community, as the App-Context grows in its capabilities, I would imagine a natural cross-fertilization of ideas would occur so that if App-Context needed any specific feature in mod_perl it would be implemented without too much trouble.

However, App-Context reaches further than just running in a mod_perl Context. App-Context is equipped with (when they are written, of course) a variety of Context classes which allow App-Context software to run in many other Contexts besides mod_perl.

What does it mean for App-Context to support Perl 5.5.3 if some of its components require a higher version of Perl?

The App-Context is like a software backplane for many services required by an enterprise application. Into that backplane plug the App-Context Services. There may be many implementations of each of the App-Context Services, making for an almost limitless array of possible combinations. (Hopefully, favorites will emerge.)

The Context is a Core Service, like Session and Config. (A Core Service is one that is not derived from App::Service but maintains the concept of "a pluggable implementation of an abstract service".) An examination of the App::Context documentation will show that there are many Contexts from which the implementer may choose to deploy the software, and the software could reasonably run on all of them. The fact that (the envisioned) App::Context::Modperl2 depends on 5.6.0 does not invalidate the fact that the App-Context depends only on 5.5.3. Implementers who choose to deploy with Services that have higher Perl version dependencies must of course satisfy those dependencies.

The important thing to realize is that there is some combination of implementations of Services which will run on 5.5.3. Thus, the implementer who is stuck with 5.5.3 on a platform does indeed have a set of possible combinations of Services which will work.

The dependency on Perl version is not the only issue like this. Some modules are dependent upon certain operating systems. It is acceptable for a Service implementation to use these OS-specific functions as long as there is some other Service implementation (perhaps lower performing) which meets the cross-platform requirement.

i.e. App::Context::Modperl2 may require 5.6.0, but App::Context::Modperl and App::Context::CGI will both run on 5.5.3.

Sometimes when we think about Enterprise Systems, we think about big budget projects with the latest new hardware. My experience with customers who are large enterprises is that they have an incredible mish-mash of systems and legacy environments. I envision that App-Context software could be installed on every platform they own (isn't Perl almost the most-ported language on the planet?). That is why I believe that App-Context must support 5.5.3 (or perhaps earlier, but I won't go there yet until I see a real need to and understand exactly what I would be giving up).