Samstag, 3. Juni 2017

Hello CoreMedia CMS 9

In any project based on the CoreMedia CMS where sites have to be created from scratch, at one point we ran into the need of the minimal content set for that site. That content area should be unrelated and unconnected to any other data and content in the system - so be really self-contained. It should make clear, which elements and settings are really needed from the wealth of options for that very project or site within the instance of the software.

Too much Content


For really some time now, CoreMedia provides a load of demo-site contents for different scenarios of the usage of their products. It is quite easy to set up a system that really shows something.

From my point of view, taking one of these demo sites and ripping out the unnecessary parts was not helpful, since you might end up - especially for new releases - with too much data that might be needed. At least a feeling of unknown dependencies sometimes remained.

URL found through Google search for "coremedia chef corp"

A small content snippet with as few elements as ever possible still seemed to be missing. And this still holds true for the latest release CoreMedia CMS-9 (and LiveContext 3 as well) as shown from a small personal survey.

Hello World


Again involved in a project based on CoreMedia products, I am now using the latest - and greatest - version of the CMS product - CoreMedia CMS 9, and asked some old friends to help me.

After showing the system and demos one of the first questions: "Ok, nice, where is the minimal Hello World site?".

From that feedback without any prior CoreMedia-related background, I assume that developers expect that part of a software - any software - to be there.

The first step now was, to ask people for their starting point content sets or Hello World sites. The result was then reworked as a CoreMedia CMS 9 Hello World site as depicted below:


It could be imported into the repository and instantly be used as a Hello World site, - and with its static content only as a Hello World site.

Configurable Content


This is a nice missing piece - I think - for CoreMedia CMS 9, but it is not really the last step for a starting point for a project. For every re-use that Hello World content has to be reworked to use the needed
  • Local
  • Path
  • Title
  • URL Segments
and the like. So there still is some work to do. But this kind of work is even more schematic and boring than the setup of the minimally needed elements.
This ended in a rewrite script and a content workspace which I now would like to make available to any interested developer in the CoreMedia-sphere.

Find the resulting code at Bitbucket, GitLab, and Github.
Enjoy. Feedback is welcome.

Dienstag, 17. Januar 2017

CoreMedia Build Docker

CoreMedia releases 7 and 8 had one very nice advantage: You didn't have to install anything else than stock Java and Maven to be able to build the workspaces.

Where we come from

This is one of the reasons, why it was possible to use standard services like travis-ci and Gitlab CI with their current container based approach to build these rather large and complex workspaces. Official docker containers for maven just did the job.

GitLab CI example to build a CoreMedia 7 workspace:
image: maven:3-jdk-7

stages:
  - build

build_modules:
  stage: build
  script:
  - mvn install -s workspace-configuration/maven-settings.xml -B

  artifacts:    paths:
    - packages/*/*/target/*.rpm
    expire_in: 3 days
    name: ${CI_BUILD_STAGE}-${CI_BUILD_REF_NAME}


cache:
  paths:
  - mvnrepo/


Where we go to

With the upcoming new CoreMedia platform for CMS and LiveContext this again changes a bit. And that bit is really that tiny, that I didn't want to lose the possibility to use the container based CI tools and go back to Jenkins (which is, what my customers provide me with).
That being said a basic docker file is prepared very quickly to include
Java 8
Maven 3
PhantomJS
The missing part is the command line tool for Sencha ExtJS, the framework the CoreMedia studio heavily relies on. This tool comes with a graphical installer and cannot be installed script based (as far as I can see).
Since I already installed that tool locally, I decided to write a little preparation script, to convert the installation to a copy-able artifact, changing an absolute symbolic link to a relative one.

Prepare a copy-able sencha Cmd copy:
S=`which sencha`
DIR=`dirname $S`
(cd build ; cp -rdp $DIR .)

SD=`(cd build/Cmd ; find -type d -name "6\.[0-9]*")|sed -e 's/^.\///g'`
(cd build/Cmd ; rm sencha-$SD ; ln -s $SD/sencha sencha-$SD)


What's left

Left out intentionally is the authentication part for maven with the CoreMedia repositories and the integration with GitLab CI, what I will be using, since these two are not really generic.

How to get it

As usual you find the sources for this small code piece on GitHub to clone, fork, and use it. And also as usual feedback is highly welcome.

Montag, 7. November 2016

Dinistiq Version 0.6

The minimalistic Dependency Injection library for Java - Dinistiq - has been published in version 0.6 some time ago.
This version still doesn't change the main outline of the project: It deals with a single - singleton - scope with a few extensions in a very small footprint library.
The library does not introduce any custom tags, but fully relies on a subset of the JSR330 definitions.
Called the qualifier release, besides dealing with qualifiers this release updates the dependencies, dinistiq relies on.

Qualifiers Introduced

Despite the fact that Dinistiq passes the TCK for JSR330, it lacked handling of qualifiers following the standard. This has been fixed with this release at the account of performance of the container.
None of my projects uses this feature, so the implementation fully relies on the TCK run integrated into the CI builds and the others dinistiq available. These tests of course have again been extended.

Availability

Starting from release 0.4 dinistiq is available from the JCenter repository at

https://jcenter.bintray.com/dinistiq/

Latest snapshots are provided at

https://oss.jfrog.org/oss-snapshot-local/ 

There are already some 0.7-SNAPSHOTs available, which reduce the number of dependencies needed by Dinistiq.

Migration

There are no special migration steps necessary for existing application.

Sonntag, 8. Mai 2016

Java 8 on OpenShift

For some time now - supported by many articles I read on the web - I thought it was a fact that

OpenShift does not support Java 8

While the OpenShift Platform as a Service environment - even some time after Java 7 left the public support track - doesn't really push you to use Java 8, it well supports it in the sense that it's simply there - but just not the default.
Of course this has been mentioned before, but there are too many recent post around stating the opposite.

OpenShift has Java 8 in Place

When you stop reading articles and just dig around in the system, you can very quickly come to the point where you see, that OpenShift has numerous Java Development Kits installed and that the default is set by Linux standard means: /etc/alternatives.
So you don't have to download it, invest space and update it regularly to the current version. This is handled by OpenShift.

Wide Choice of JDKs

The /usr/lib/jvm folder has everything you might need - JDK-wise - and a simple scriptlet in your deployment will bring you to the needed JDK:

JAVA_VERSION=1.8.0
export JAVA_HOME=`find /usr/lib/jvm -name "$JAVA_VERSION"`
if [ -z JAVA_HOME ] ; then
  echo "INTENDED JAVA VERSION NOT FOUND!"
fi 

So, no need to stay away from Java 8 because of something related to OpenShift (anymore).

Prepared Quickstart for Java 8, Gradle, and Tomcat

For the users of Gradle and Tomcat I updated this in my quickstart on github.com based on the DIY cartridge.

Freitag, 6. Mai 2016

Tangram Release 1.1

The 1.1 release is a consolidation release removing outdated stuff leaving
room for latest versions of the used libraries and modules and providing some
polishing in the existing feature set.

Beyond the App Engine

The sources and release notes can be found on github.com, the binaries again reside on JCenter. This release includes some notable overdue dependency updates like
  • Servlet 3.1 API now included
  • JSP 2.1 now included
  • DataNucleus updated to 4.1
  • dinistiq updated to 0.5
To allow this, it leaves out the Google App Engine support.

Gretty Plugin

Since not only the Google App Engine has a defect with old releases of Java
Web APIs but also Gradle has a obsolete Jetty plugin with an old version of
Jetty, all examples have been migrated to Gretty which is now the preferred container integration and even application packaging option.

Morphia Support

Latest Tangram applications always used Mongo DB as the storage backend with either JPA or JDO as the API. If you don't need the API abstraction or want an even leaner storage integration (before that DataNucleus JDO and EBean had the smallest footprint) you now can start using the "Mongo DB only Mapping" Morphia.

Markdown Support

Up to now every property using char[] for getters and setters was handled as structured text. It was edited as an HTML fragment (p or div) using the CKEditor.
Starting with this release, every property using org.tangram.content.Markdown for the getters and setters is handled through the CodeMirror editor as Markdown syntax. It is transformed to HTML in rendering.
Also templates can now be in Markdown syntax whenever appropriate while also here the output is transformed to HTML before passing the templates to the other layers of Tangram.

Taking Care of the Modification Date and Time

If you need to track the modification time of an object through the Tangram editor module or the ftp service, you can now do so and the standard storage backed classes of Tangram now do so.

File Restart Cache

The restart caches used for faster restarting of the application with data which should not be changing between sessions of the application now only have to be cleared manually on error conditions but not on changes of the application any more.

Exporting and Importing

The export and import features of Tangram are now strictly symmetrical: Full content export and import  - more or less independent of the storage flavour in use - and a code exporter and importer for the sources stored in the repository.

Fat JARs

Additionally it is now possible to leave out the "war" dependencies since
resources like templates can be placed in the respective JARs.

Extended testing

Tangram now for the first time has a decent test coverage. While there still
is room for more in depth testing of many functions, we now can put more trust
into the single snapshot builds. This was quite important since most Tangram
applications not residing on the Google App Engine already migrated to the 1.1
level.

Sonntag, 22. November 2015

Tangram Release 1.0

Since Tangram reached more than the set of features originally intended back in 2009, it was time to call the current state a 1.0 release.

Why?

The name Tangram points out that the main objective was to create web applications from a fixed set of modules and options which can form nearly any shape you would need. You can start quickly with a limited set of functionality and let the application grow. Fixed set in this case only means that you will need e.g. a model implementation but still have a choice, which one you would want to use. Also the glueing together with a Dependency Injections solution provides several options.
Another intention was to provide reasonable defaults for nearly every aspect where this is possible. You don't need to copy tons of files which do things technically necessary, which you didn't think of at that time. You only set up the things you know that you need them, and all the other features stay quiet. At least they should do something reasonable without getting in your way.

Users

Tangram is in production use since 2011 by Provocon, Ponton, Naturinspiriert.org and others.

Dynamic Extendibility

For many extensions you will not even need a deployment but a change in the data repository forming small and stable cores where the application as whole can be modified to changing requirements quickly, easily and securely.
Apart from the obvious presentation stuff this includes Groovy codes to create and parse URL, extend the business logic, and even the model layer with new model classes.

Core Features

We provide object oriented templating for objects from JPA, JDO, EBean, and CoreMedia repositories. The core system emphasises dynamic web programming through CSS, JavaScript, Velocity Templates, and Groovy Codes in the repository together with basic CDN support, code minification, and caching support.

URLs can be formatted in nice, SEO friendly schemes and be easily mapped to actions to provide the result views for users.
The implementation of authentication and authorization now uses the pac4j set of libraries to provide a seamless and easy integration of plain user lists in files or the repository, OAuth providers, OpenID providers, Google App Engine user service and others for the web application and the base system itself.
This authentication solution is used grant access to the system itself but also to provide support for protected content areas in your applications. You only have to focus on the question, which content needs to be protected by a access granting scheme. Logins can be a generic login page or integrated in you application design.

Generic Editor

Except for CoreMedia we provide a generic web editor which is now responsive (there is now separate mobile editor anymore) which can be loaded from the respective cloud locations of the used components (CKEditor, CodeMirror). Also it can be extended with our dynamic web programming facilities.

Glueing Stuff Together

The mentioned default set of configurations and the option to customize this to your needs is achieved by Dependency Injection libraries. Tangram supports the usage of the Spring Framework, Google Guice, and dinistiq. It should be possible to add other solutions as well provided that they offer a decent set of functionality (which is not always the case for smaller DI libs.) including e.g. optional values, overriding of configurations, and deep generics introspection.

Readware

To illustrate the usage and provide a nice starting point, a set of example applications is provided in sync with the releases.
A documentation wiki is now starting.

Where the 1.0 Release Ends

One real limitation is the older set of JSP/Servlet APIs which need to stay in place to support the Google App Engine. So this is the final release to support the App Engine to be able to move ahead to newer APIs for several areas. This will not be achieved in the 1.0 branch.

Mittwoch, 28. Oktober 2015

Gradle Plugin for EBean, JPA, and JDO Enhancing along with Minification and Overlaying

During the development of the Tangram framework project a set of build related things went into a plugin for the Gradle tool.
The functionality of this Tangram Gradle Plugin is only in very small parts directly related to Tangram. It is more or less a general purpose plugin for applications needing
  • Byte-code transformation of model classes for
    JDO, JPA, and Ebean ORM layers
  • Minification of CSS and JavaScript codes to be placed in WAR artifacts
  • Support underlying of WAR files into others (similar to overlays)
The good news for today is, that with the latest version 1.0.5 the plugin can be used from the central Gradle plugins repository with the - not very surprising - id tangram.gradle.plugin. So some of the usage notes have to be aligned with this situation.

Usage

Just a few lines have to be added to your Gradle build script to use the plugin:

plugins {
  id "tangram.gradle.plugin" version "1.0.5"
}

All of the following steps described here take place without any additional configuration.

Prepare EBean, JPA, and JDO Model Classes

When used with Java projects - and when some data model classes are discovered, - the plugin tries to prepare them for use the respective Object Relational Mapper (ORM). The ORM APIs supported are
These APIs in turn are supported by a number of implementations. The supported implementations are
These OR-Mapper API implementations require (DataNucleus and EBean) or recommend (the others) to apply byte-code transformations called "Enhancing" or "Weaving" to the class files. The compiled code is extended with some database access support to implement the active record pattern more or less seamless.
The API and implementation library in use is discovered from the names of the elements of the class path of the project. If one of the mentioned libraries is found, the corresponding byte-code transformation is applied to the appropriate step (post compile or pre jar creation).

martin@nelson:~/proj/tangram/sites/naturinspiriert$ gradle clean build
:clean
:compileJava
Performing DataNucleus JPA byte code transformation.
ENHANCED (Persistable) : org.naturinspiriert.RootGroup
ENHANCED (Persistable) : org.naturinspiriert.Topic
ENHANCED (Persistable) : org.naturinspiriert.AbstractGroup
ENHANCED (Persistable) : org.naturinspiriert.ImageData
ENHANCED (Persistable) : org.naturinspiriert.Article
ENHANCED (Persistable) : org.naturinspiriert.Linkable
ENHANCED (Persistable) : org.naturinspiriert.Group
DataNucleus Enhancer completed with success for 7 classes. Timings : input=60 ms, enhance=54 ms, total=114 ms. Consult the log for full details
7 classes enhanced.
:processResources
:classes
:jar
:war

:assemble
:compileTestJava UP-TO-DATE
:processTestResources UP-TO-DATE
:testClasses UP-TO-DATE
:test UP-TO-DATE
:check UP-TO-DATE
:build

BUILD SUCCESSFUL

Total time: 4.53 secs

(Highlighted message to indicate use of the DataNucleus Enhancer)

The byte-code transformations directly use the transformer of the respective library in use except for the OpenJPA case, where the ant task of the enhancer is integrated. Some of the transformers issue some logging.

Switching Off

Additionally it is possible to switch of the byte code transformation for JPA annotated classes by adding to your build file

// build.gradle: 
enhancer.enabled=false

in case this might be necessary e.g. to only use the other parts of the plugin. Also EclipseLink, Hibernate, and OpenJPA support the use of plain Java classes without the byte-code transformation.

JavaScript and CSS Minification

When used in conjunction with the war plugin, CSS and JavaScript resources are automatically minified.
The plugin checks for resources with a filename extension .css for Cascading Style Sheets and .js for JavaScript. Matching resources are minified using the YUI Compressor.
It is not possible to minify resource included from archive files but only file resources local to your project. So contents from archives - while being included in the resulting web archive - cannot be minified. (We would expect WAR files to contain minified resources like WAR files generated using this plugin do.)

Web Application Underlying

The plugin introduces a configuration named webapp for modules using the war plugin. Dependencies given for this configuration are extracted into the resulting war artifact.

// build.gradle:
dependencies {
  webapp "tangram:tangram-core:$tangram_version:war@war"
  webapp "tangram:tangram-editor:$tangram_version:war@war"


  compile "tangram:tangram-core:$tangram_version"
  // Persistence API JPA
  compile "tangram:tangram-jpa:$tangram_version:nucleus"
  compile "org.datanucleus:datanucleus-api-jpa:$versions.datanucleus"
  compile "org.datanucleus:datanucleus-core:$versions.datanucleus"
  compile "$versions.jdo_api"
  compile "$versions.persistence_api"
  runtime "org.datanucleus:datanucleus-mongodb:$versions.datanucleus"

  compile "tangram:tangram-editor:$tangram_version"
  runtime "tangram:tangram-dinistiq:$tangram_version"
  runtime "org.slf4j:slf4j-log4j12:$versions.slf4j"


  providedCompile "$versions.servlet_api"
  providedCompile "$versions.jsp_api"
}


This process is not really described well if called overlay so I call in underlying, since your web application's directory in fact is the overlay so the other archives referenced and included must be an underlying.
If your WAR relies on the contents of another pre-packaged or incomplete WAR, the contents of the latter will be copied into your resulting web application while you can override any file in this archive from your local web application contents directory.

Version List

The plugin introduces a version object which collects some version strings for a number of libraries. This ensures that any project using the plugin can use these libraries with recent versions and version changes are applied in sync. The use of this part is optional and you have to explicitly use the versions in your build file since this cannot be applied transparently.
Some random examples:

dependencies {
  compile "org.pac4j:pac4j-openid:$versions.pac4j"

  compile ("org.apache.openjpa:openjpa:$versions.openjpa") {
    exclude group: 'asm'
  }

  compile "org.eclipse.persistence:org.eclipse.persistence.jpa:$versions.eclipselink"
  compile "$versions.persistence_api"

  compile "org.hibernate:hibernate-core:$versions.hibernate"
   
  compile "org.datanucleus:datanucleus-api-jpa:$versions.datanucleus"
  compile "org.dataucleus:datanucleus-core:$versions.datanucleus"
  compile "$versions.jdo_api"
  runtime "org.slf4j:slf4j-log4j12:$versions.slf4j"

  testCompile "org.testng:testng:$versions.testng"

  // your container will have this for you
  providedCompile "$versions.servlet_api"
  providedCompile "$versions.jsp_api"
}


Manual Mode

Of course it is still possible to call the methods performing the different tasks directly like described in the 0.9 plugin blog post. This should only be necessary if you e.g. want to enhance files in the unit test section of your code, which is considered a very rare case.

Dienstag, 27. Oktober 2015

Dinistiq Version 0.5

The minimalistic Dependency Injection library for Java - Dinistiq - has been published in version 0.5.
This version still doesn't change the main outline of the project: It deals with a single - singleton - scope with a few extensions in a very small footprint library. The library does not introduce any custom tags but fully relies on a subset of the JSR330 definitions.
Despite its limited scope it still successfully executes the TCK and comes with a reasonable unit test coverage. These unit tests have been migrated to TestNG to be in sync with my other projects reflecting the license which better fits into the set of licenses of the used dependencies in this project.

Fixes

The once more extended test coverage after the migration to TestNG showed two bugs relating to class discovery. These have been fixed in this release.

Leaner Web Integration

Due to the use of older versions of the Servlet API, dinistiq had to use a central dispatcher servlet dispatching to other so called registerable servlets. This level of indirection has been complete removed. With dinistiq 0.5 the web integration is mostly rewritten now depending on the Servlet API version 3.1.
Sadly this also removes its URL pattern matching capabilities.
RegisterableServlet instances from the dinistiq scope are now registered with the Servlet container directly. They now present a number of Servlet specification compatible URL patterns (no regular expressions any more - sorry) and and the same ordering indicator as with previous releases.

Availability

Starting from release 0.4 dinistiq is available from the JCenter repository at

https://jcenter.bintray.com/dinistiq/

Latest snapshots are provided at

https://oss.jfrog.org/oss-snapshot-local/ 

There are already some 0.6-SNAPSHOTs available, but they don't present changes worth mentioning right now.

Migration

Take a look at your web.xml. If you see something like

  <servlet>
    <servlet-name>dinistiq</servlet-name>
    <servlet-class>dinistiq.web.DinistiqServlet</servlet-class>
  </servlet>
  <servlet-mapping>
    <servlet-name>dinistiq</servlet-name>
    <url-pattern>/s/*</url-pattern>
  </servlet-mapping>


simply remove it!
The RegisterableServlet interface has been change as mentioned. You will have to update you regular expressions to Servlet specification compatible URL patterns. Of your the respective method signature was changed from

    @Override
    public Set<String> getUriRegex() {
        Set<String> result = new HashSet<>();
        result.add("");
        result.add("/.*");
        return result;
    } // getUriRegex()

to

    @Override
    public Set<String> getUrlPatterns() {
        Set<String> result = new HashSet<>();
        result.add(source.getDispatcherPath()+"/");

        result.add(source.getDispatcherPath()+"/*");
        return result;
    } // getUrlPatterns()

Mind the missing regular expression syntax and that the full URI path has to be given.


Samstag, 3. Oktober 2015

Gradle Plugin for JDO, JPA, Ebean etc in Tangram grown up

As part of the Tangram framework project a set of build related things went into a plugin for the Gradle build tool.
Only in part this plugin is directly related to Tangram but offers some generic tools suitable for many projects.
The plugin does
  • Support underlying of WAR files into others (similar to overlays)
  • Minify CSS and JavaScript codes to be placed in WAR artifacts
  • Byte-code transform model classes for JDO, JPA, and Ebean ORM layers
Up to version 0.9 the plugin was more or less a collection of utility methods. It is now redesigned to do the same job transparently when appropriate and at the appropriate stage of the build process.

Usage

Just a few lines have to be added to your Gradle build script to use these add on functionality:

buildscript {
  repositories {
    jcenter()
  }
  dependencies {
    classpath 'tangram:gradle-plugin:1.0.0'
  }
}


apply plugin: 'tangram'

All of the following steps described here take place without any additional configuration.

Web application underlying

The plugin introduces a configuration named webapp for modules using the war plugin. Dependencies given for this configuration are extracted into the resulting war artifact.
This process is not really described well if called overlay so I call in underlying, since your web application's directory in fact is the overlay so the other archives referenced and included must be an underlying.
If your WAR relies on the contents of another pre-packaged or incomplete WAR, the contents of the latter will be copied into your resulting web application while you can override any file in this archive from your local web application contents directory.

Minification

When - again - used in a project also using the war plugin, all files included from the project directory will be checked for CSS and JavaScript file extensions. Matching files are minified using the YUI Compressor.
It is not possible to minify resource included from archive files. So contents from archives while being included in the resulting WAR cannot be minified. (We would expect WAR files to contain minified resources like WAR files generated using this plugin do.)

Handle Model Classes

When used with Java projects and when some data model classes are discovered, the plugin tries to prepare them for use with one of the ORM frameworks Java Data Objects (JDO), Java Persistence API (JPA), or Ebean.
The OR-Mappers require or recommend to apply byte-code transformations called "Enhancing" or "Weaving" to the class files for the data model related classes.
The plugin supports DataNucleus Access Platform (JDO and JPA), EBean, EclipseLink, Hibernate, and OpenJPA. All the calling options from the 0.9 version of the plugin have be removed and the APIs are auto-discovered so the whole process is now - except for logging output - transparent during build.

Version List

The plugin introduces a version object which collects some version strings for a number of libraries. This ensures that any project using the plugin can use these libraries with recent versions and version changes are applied in sync. The use of this part is optional and you have to explicitly use the versions in your build file since this cannot be applied transparently.
Some random examples:
dependencies {
  compile "org.pac4j:pac4j-openid:$versions.pac4j"

  compile ("org.apache.openjpa:openjpa:$versions.openjpa") {
    exclude group: 'asm'
  }

  compile "org.eclipse.persistence:org.eclipse.persistence.jpa:$versions.eclipselink"
  compile "$versions.persistence_api"

  compile "org.hibernate:hibernate-core:$versions.hibernate"
   
  compile "org.datanucleus:datanucleus-api-jpa:$versions.datanucleus"
  compile "org.dataucleus:datanucleus-core:$versions.datanucleus"
  compile "$versions.jdo_api"
  runtime "org.slf4j:slf4j-log4j12:$versions.slf4j"

  testCompile "junit:junit:$versions.junit"

  // your container will have this for you
  providedCompile "$versions.servlet_api"
  providedCompile "$versions.jsp_api"
}

Manual Mode

Of course it is still possible to call the methods performing the different tasks directly like described in the 0.9 plugin blog post.
Also it is possible to switch of the byte code transformation for JPA annotated classes by adding to your build file

// build.gradle: 
enhancer.enabled=false

in case this might be necessary e.g. to only use the other parts of the plugin.

Sonntag, 26. April 2015

Stock Apache Tomcat and Gradle on OpenShift

For my ongoing work on Tangram a wanted to use the RedHat OpenShift platform. In this post I'm going to outline the generic part of this which now is extracted as a Quickstart on github via https://github.com/mgoellnitz/openshift-tomcat-gradle-quickstart.
The requirements where the following: Build must be possible with Gradle and a plain Apache Tomcat webcontainer - not a JBoss Tomcat derivative - is needed. For the time being I still left out the option to use Jenkins, since the way to re-deploy on git-push OpenShift gives me, is quite usable.
So this results in the idea of a combination of the work of Shekhar Gulati at https://www.openshift.com/blogs/run-gradle-builds-on-openshift with the code at https://github.com/shekhargulati/gradle-openshift-quickstart and the tomcat Quickstart https://github.com/openshift-quickstart/openshift-tomcat-quickstart
Looking at the changes needed to use Gradle, I decided to integrate them manually and modify them along the way. Also the Tomcat Quickstart will not be used directly but serves as a template, since I don't want to pull the Apache Tomcat binaries into the git repository. As opposed to many other examples I came across, the installation of the RedHat cloud command line tools is completely left out. You might want to use them, but they are not needed for this example.
First of all I had to create an application (named "test") and had to take "DIY 0.1" as the type of application and clone the resulting repository.

git clone ssh://524e65...094@test-application.rhcloud.com/~/git/test.git/
cd test


This repository resembles the working directories of your application on OpenShift and is not just a development set of folders. So you will find some scripts here to interact with the infrastructure of OpenShift. These scripts have to be customized to fetch and configure Apache Tomcat and Gradle.
Now it's time to add script element to fetch the tool.

# .openshift/action_hooks/build:

# gradle build
set -x

DIR=`pwd`

# obtain tomcat
export TOMCAT_VERSION=7.0.61
if [ ! -f $OPENSHIFT_DATA_DIR/apache-tomcat-${TOMCAT_VERSION}.tar.gz ] ; then
  wget http://apache.../v${TOMCAT_VERSION}/bin/apache-tomcat-${TOMCAT_VERSION}.tar.gz
fi
if [ -f $OPENSHIFT_DATA_DIR/apache-tomcat-${TOMCAT_VERSION}.tar.gz ] ; then
  rm -rf tomcat
  tar xvzf $OPENSHIFT_DATA_DIR/apache-tomcat-${TOMCAT_VERSION}.tar.gz
  mv apache-tomcat-${TOMCAT_VERSION} tomcat
  cd $OPENSHIFT_DATA_DIR/tomcat
  rm -rf logs
  ln -s $OPENSHIFT_LOG_DIR logs
  cd $OPENSHIFT_DATA_DIR
fi

# obtain gradle
export GRADLE_VERSION=2.2.1
if [ ! -d $OPENSHIFT_DATA_DIR/gradle-${GRADLE_VERSION} ] ; then
  cd $OPENSHIFT_DATA_DIR
  wget http://services.gradle.org/distributions/gradle-${GRADLE_VERSION}-bin.zip
  unzip gradle-${GRADLE_VERSION}-bin.zip
  rm -f gradle-${GRADLE_VERSION}-bin.zip
fi

cd $DIR

cd $OPENSHIFT_REPO_DIR
export GRADLE_USER_HOME=${OPENSHIFT_DATA_DIR}gradle
export GRADLE_HOME=${OPENSHIFT_DATA_DIR}gradle-${GRADLE_VERSION}
export PATH=$GRADLE_HOME/bin:$PATH
if [ -f openshift-build.gradle ] ; then
 gradle --stacktrace -b build.gradle clean build
fi


Use at least version 2.2 of Gradle due to http://issues.gradle.org/browse/GRADLE-2871. Also Gradle needs some local directory on the target host.


# .openshift/action_hooks/pre_build:
 

# gradle storage dir build
set -x

if [ ! -d $OPENSHIFT_DATA_DIR/gradle ] ; then
  mkdir $OPENSHIFT_DATA_DIR/gradle
fi


The default stop action hook script contains some references on a ruby server and is missing a nice tomcat handling.


# .openshift/action_hooks/stop:
source $OPENSHIFT_CARTRIDGE_SDK_BASH

# The logic to stop your application should be put in this script.
if [ -z "$(ps -efww | grep tomcat | grep -v grep)" ] ; then
  client_result "Tomcat is already stopped"
else
  if [ -x $OPENSHIFT_DATA_DIR/tomcat/bin/shutdown.sh ] ; then
    $OPENSHIFT_DATA_DIR/tomcat/bin/shutdown.sh
  else
    kill `ps -efww | grep tomcat | grep -v grep | awk '{ print $2 }'` > /dev/null 2>&1
  fi
fi
exit 0



For safety reasons it might be a good idea to delete unexploded wars from the webapp directory of the tomcat installation


# .openshift/action_hooks/deploy:
set -x

# Addition:
for i in `ls $OPENSHIFT_DATA_DIR/tomcat/webapps/*.war`; do
  dn=`basename $i .war`
  echo "removing $dn"
  rm -rf $OPENSHIFT_DATA_DIR/tomcat/webapps/$dn
done


The start script has to be modified not to use ruby but the tomcat installation


# .openshift/action_hooks/start:
set -x
cp -rdp ${OPENSHIFT_REPO_DIR}/tomcat/conf/* ${OPENSHIFT_DATA_DIR}/tomcat/conf/
cp -rdp ${OPENSHIFT_REPO_DIR}/tomcat/lib/* ${OPENSHIFT_DATA_DIR}/tomcat/lib/
cp -rdp ${OPENSHIFT_REPO_DIR}/tomcat/webapps/* ${OPENSHIFT_DATA_DIR}/tomcat/webapps/
cd $OPENSHIFT_DATA_DIR/tomcat
sed -ig 's/OPENSHIFT_DIY_IP/'$OPENSHIFT_DIY_IP'/g' conf/server.xml
sed -ig 's/OPENSHIFT_APP_DNS/'$OPENSHIFT_APP_DNS'/' conf/server.xml
bin/startup.sh


After these preparation steps it would be time to integrate your application by providing a build.gradle script in the root path and all the needed source files.
When using a custom domain, let this domain point to your Applications hostname via a DNS CNAME record. Then create an alias with your domain name for the application using the RedHat Cloud command line tool. (Or create the alias via the console web application) Don't forget to add this domain name to your server.xml override.

Mittwoch, 25. März 2015

Dinistiq Version 0.4

The minimalistic Dependency Injection library for Java Dinistiq has been published in version 0.4.
This version doesn't present anything specifically new. It still deals with a single - singleton - scope with a few extensions in a very small footprint library. The library does not introduce any custom tags but fully relies on an subset of the JSR330 definitions.
Despite its limited scope it still successfully executes the TCK and comes with a once more extended unit test coverage.

Availability Release

One of the main changes of this release is the availability of the release artifacts through the JCenter service of bintray. This renders the integration of Dinistiq even easier since you will most likely not need to add another repository or mirror to your infrastructure.
Dinistiq is in daily production use for some time now and will remain supported as long as there is not always a need for more feature complete solutions like Google Guice or the Springframework.

Next Steps

The versioneye service shows that Dinistiq is up to date with its dependencies with only one intentional exception. This release will be the last to support the Google App Engine through the use of outdated Servlet API versions.

Freitag, 5. Dezember 2014

Dinistiq Version 0.3

Dinistiq ist entstanden, weil es schneller zu implemtieren war, als die Integration anderer DI-Framework und Container in Tangram dauerte.
Da es dann eine Weile einfach alles tat, was notwendig war, wurde nur geringe Code-Optimierungen eingeführt.

Qualitätssicherung

Zu seinem ersten Geburtstag hat Dinistiq aber ein echte Upgrade verdient.
Seit August befindet es sich im produktiven Einsatz in einem Testframework zur Validierung der Einhaltung der Anforderungen für ein großes deutsches Medienportal, das auf einer neuen technischen Basis komplett reimplementiert wird.
Dabei sollte ein einfacher, effizienter, wartbarer und schlanker Technologiestack zu Einsatz kommen.
Die alternativen Stapelten Selenium, PHP und Shellskripte und waren nicht in der Lage, die angeforderten Varianten an Plattformen und Browsern mit Testfällen zusammenzuführen. Außerdem paßten sie natürlich nicht in die reine Java-Welt des Restes des Projektes.
Gleichzeitig sollten veraltete Techniken wie Maven natürlich ebenso außen vor bleiben.
Nun muß sich Dinistiq jeden Tag darum kümmern, hunderte von Testfällen mit jeweils unterschiedlichen Browsern und Zielplattformen durch Dependency Injection zusammenzuführen.

Qualitätssicherung

Dinistiq befindet sich nun täglich im Projekteinsatz und es ist nicht mehr hinnehmbar, wenn dabei genutzte Funktionen nicht zuverlässig nutzbar sind. Da Dinistiq mit dem Einsatz reift, sind dennoch häufige Snapshots nötig gewesen, sodaß ist einer verstärkten UnitTest Abdeckung und dem Einsatz von PMD zur Source-Code Kontrolle zwar nur ein Fehler befunden wurde, das Neu-Einführen von Fehlern aber von vorneherein úmgangen wurde.
Zur Generierung der Übersicht wird Jacoco eingesetzt.
Der Impuls war aber hier auch, daß der Einsatz in der Qualitätssicherung für ein ungleich größeres Projekt erfolgen sollte und so mußte sich Dinistiq als Testfeld für die angeforderten Qualitätsmaße an den Source-Code zur Verfügung stellen.
Bisher war dieser Schritt ein voller Erfolg und in jedem Falle viel einfacher umzusetzen als alle mir bekannten Alternativen wie Springframework, Google Guice, Weld, OpenWebBeans, die den Job sicherlich auch erledigen können.
Dafür gibt es heute den Stempel "0.3".

Details

Mit Version 0.3 ist auch das Standard Repository für die Artefakte gewandert und der CI Server ebenfalls. Dies ist nur der Tatsache geschuldet, daß Cloudbees sich aus dem Geschäft zum Jahresende zurückzieht.

Samstag, 30. August 2014

Tangram CoMA erwacht

Da ich nach einem langen Sommeurlaub nun wieder schwerpunktmäßig mit dem CoreMedia CMS zutun habe, habe ich mir auch im Tangram nach sehr langer Ruhe mal wieder die entsprechende Anbindung angesehen.

Ein altes Beispiele

Seit geraumer Zeit gibt es eine Beispielanwendung zum CoreMedia Adapter CoMA im Tangram Framework. Diese litt aber daran, daß das Aufsetzen der Umgebung dazu nicht ganz trivial war. Alleine schon deshalb konnte ich nicht soviel Energie in die Weiterentwicklung stecken.
Für diese Anwendung stehen Möglichkeiten von Tangram, Content einzugeben, nicht zur Verfügung, da der CoMA nur lesend arbeitet. Um dennoch Inhalte und ein paar Templates zum Zeigen zu bekommen, nutze ich die uralte Beispielanwendung MenuSite von CoreMedia selbst.
Man benötigt also zum Nutzen der Anwendung eine Datenbank, wie sie ein CoreMedia Content Server für die MenuSite angelegt und mit Daten befüllt hätte.
Bisher waren die Schritte, die nötig waren, um diese Datenbank aufzusetzen nur zur Ausführung beschrieben. Alle andere Beispiele kann man einfach bauen und starten.

Neue Server

Meine Experimente mit Gradle zum Zusammenbau von CoreMedia Softwarekomponenten ohne das bei mir nicht sehr beliebte Maven haben allerdings nun einen Stand erreicht, in dem man Skripte und kleine Workspaces präsentieren kann, mit denen man die benötigte Datenbank erstellen kann.
Die Content Management Server Webanwendung unter
https://github.com/mgoellnitz/cm-cms-webapp
und die Content Management Server Tools unter
https://github.com/mgoellnitz/cm-cms-tools
Die harte Nuß dabei war weniger der Server als das Tools Paket. Aber ohne die Tools konnte ich die Daten natürlich nicht in den Server importieren. Außerdem hat das von mir gewählte Tomcat Plugin für Gradle sich nicht von der freundlichen Seite gezeigt. Mittlerweile weiß ich, daß auch andere Anwendungen als der CoreMedia Content Server unter dem falsch zusammengestellten Classpath leiden.

Und alles ohne Maven

Die genaus Beschreibung findet sich im Beispiel Repository zu Tangram.

Sonntag, 17. August 2014

Tangram Release 0.9

Over the last year, Tangram has changed very much and does not look the same in all of its modules and options.

Persistence Options

The Tangram Dynamic Web Application Framework has been extended to support the Java Persistence API (JPA) and EBean as persistence layers in addition to the already available Java Data Objects solution. This greatly extends the number of options for platforms to use with Tangram.
The Tangram examples reflect this with the JDO/DataNucleus, JDO/Google App Engine, EBean, and JPA example where the later is switchable with tested API implementations for OpenJPA, EclipseLink, DataNucleus, and Hibernate.
The simple and more or less generic JDO editor now is a generic editor for all the storage solutions and built as a separate module. Except for Google App Engine this module can be left out if there is another solution available to get the content into the repository.

Dynamic Model Extenions

In addition to the Groovy implemented classes stored in the repository, which are used to extend the view and controller part of the application, you can now create JDO annotated classes which are immetiately usable as model classes in the repository. This option can only be presented for the JDO implementation.
To make the building of Tangram itself, any Tangram application, or even any application using byte-code transformation for JDO, JPA, and Ebean easier, a gradle plugin has been introduced. The examples and over blog entries illustrate the usage of this plugin.

IDE Integration

To be able to synchronize codes stored in the repository a separate FTP module has be introduced. Most IDE due to their strong PHP past support FTP better than any other protocol. CSS, JavaScript, Groovy codes can be imported and exported.

Generic Import and Export

Additionally and apart from the FTP module Tangram now contains a generic importer and exporter for the whole content stored in the repository. The XML representation can be used to transfer content between application using JPA, JDO, or EBean. IDs cannot be preserved.

Framework Independency

For the whole Tangram code the Springframework now is only an option to do the setup of the application by Spring's Dependency Injection IoC-Container and for the controller and view parts of the application.
Along this way it was easier to present a custom DI component, Dinistiq, than to use one of the available solutions. Tests have been done with Google Guice, TinyDI, and JSR-330. As of Tangram 0.9 none of these are more than experiments. Work on a Tangram JavaEE integration is also in an experimental stage and would require a major refactoring of some parts of Tangram, see other entries in this blog on the JavaEE topic.
It should now be fairly easy to do more experiments with other IoC/DI frameworks and containers or to integrate other web frameworks starting from the new plain servlet based solution.

Options

Tangram 0.9 has been tested, verified, or is even in production use on the following platforms: Google App Engine, run@CloudBees, OpenShift, and Standalone with Apache Tomcat using RDBMS or MongoDB.
The option matrix now looks like this:
Persistence: JDO, JPA, EBean
IoC: Springframework, Dinistiq
Security: Spring Security, Apache Shiro
Hosting: On Premise, OpenShift, Cloudbees, Google App Engine
Storage: RDBMS, No-SQL e.g. MongoDB, Files

Code Reduction

Tangram started as an idea to plug existing components together to form a web application with as little glue code as possible. Some of the ideas like object oriented templating though had to be coded explicitly. After this starting point the Tangram codes started growing and the 0.9 release got an additional design goal to stop this. Still some of the codes in Tangram duplicate functionality which would otherwise externally be available but avoids a depedency just to call one method. And there are still codes duplicates within Tangram where the different options e.g. for persistence might get a stronger common code base. In these areas it is still not clear if they will differentiate or grow together.

Outlook

The experiences with CapeDwarf showed the way into a Java EE module as an alternative to dinistiq, the Springframework, and the many tests with e.g. Google Guice. The not that encouraging experiences with the EBean ORM might lead to discontinued support for that module. Dinistiq will become more Java EE / CDI compatible and CDI without the full Java EE stack also seems to be an interesting option. Work on the generic import and export functions using XStream needs to be extended and easier integration of OpenID or OAuth outside of the Google App Engine is needed for many applications. Google App Engine with JPA is still on the list but not desperately needed.

Donnerstag, 7. August 2014

Dinistiq 0.2 Release

The minimalistic Dependency Injection solution for the setup of component based Java applications using Singletons and the JSR330 annotations has stabilized to a 0.2 release.
With some minor bug fixes it also better deals with misconfigured setups.
All system properties from the JVM are now part of the scope. The logging framework has been changed from Apache Commons Logging to Simple Logging Facade for Java (slf4j).
Decent handling of maps and booleans has been added and circular dependencies are now detected and reported.
Along all these minor changes the documentation has been extended and a javadoc archive is added to the deliverables.

Freitag, 25. Juli 2014

150 Lines just for one Collection

Some 15 years ago I left the Java Enterprise Edition train and am now pushed by things like CapeDwarf on JEE Servers to the evaluation of running Tangram somehow in a JEE environment. I read the promise, that things have become a lot easier and that JEE has adopted the annotation and convention based style I now feel familiar.
I came across the JSR330 annotations from the javax.injection package which were introduced in the JEE world. JSR330 Annotations are recognized by several Dependency Injection frameworks like Guice, Springframework, my homegrown dinistiq, and obviously the JEE Containers.

Step 1: Get rid of Spring

This is were I started to make Tangram less depending on the Springframework directly or provide interfaces to be implemented in a spring and a non-spring way. For the dependency injection part it was mostly as easy as to migrate from

    @Autowired
    private ClassRepository classRepository;


to

    @Inject
    private ClassRepository classRepository;

I lost the required = false option at that time but was able to live with that given the advantages of portability.

Step 2: Find Alternatives

The spring parts of the application where re-implemented using a plain servlet based solution, and springsecurity was replaced by Apache Shiro for those scenarios.
After having migrated to the new annotations most of my Components where portable across containers. So I don't have to sell my soul to a special DI framework for very many cases, which helps choosing the right infrastructure for every project. But there are several details which still remain problematic, most notably the missing required = false option and the missing collection of instances to be able to automatically let the whole set of instances be injected.

Step 3: Annoying Details

The annoying collection problem looks like this:

@Autowired(required = false)
private Collection<ControllerHook> controllerHooks = new HashSet<ControllerHook>();


Later I migrated this to the JSR-330 Annotations - which also works with spring.

@Inject
private Collection<ControllerHook> controllerHooks = new HashSet<ControllerHook>();


I made this work with dinistiq as well.
With Guice I learned, that it didn't like to collect the instances for me and provide me with the collection out of the box.
I'm expecting here, that any instance implementing ControllerHook from the application context gets added to a collection which then is injected into the consuming component. Spring does it, dinistiq does it, Guice can be convinced to do it by the additional Multibinder:
 
Multibinder<ControllerHook> multibinder = 
                            Multibinder.newSetBinder(binder(), ControllerHook.class);
multibinder.addBinding().to(UniqueUrlHook.class);
multibinder.addBinding().to(ProtectionHook.class);

Step 4: Show Stopper

Inspired by the CapeDwarf project, promising to be able to deploy applications developed for the Google App Engines and their APIs on a JEE infrastructure, I tried to deploy GAE based Application to such an Infrastructure based on a JBoss AS7 or Wildfly 8 Server using CapeDwarf Versions 1 or 2 respectively.
This approach doesn't work for Tangram since the used DI solution within the application - be it dinistiq or the Springframework - interferes with the JEE Container's CDI implementation. The Application Server starts to interpret the annotations intended for the other DI framework. This would instantiate the application components twice and right at the moment it also fails on some of those components.

Step 5: JEE Wordiness

So quite naturally I now went over to directly migrate Tangram to the JEE world. But also JEE has the same problem as plain Guice has.

@Inject
private Collection<ControllerHook> controllerHooks = new HashSet<ControllerHook>();

This time the solution really is ugly. It seems I have to implement a different consuming component and thus instead of the two lines, a separate ControllerHookProvider is needed

public interface ControllerHookProvider {

    Collection<ControllerHook> getControllerHooks();

} // ControllerHookProvider

Of course the generic implementation looks like the good old two lines which did the whole job for me so far.
 
public class GenericControllerHookProvider implements ControllerHookProvider {
    @Inject
    private Collection<ControllerHook> controllerHooks;


    public Collection<ControllerHook> getControllerHooks() {
        return controllerHooks;
    } // getControllerHooks()

} // GenericControllerHookProvider

For JEE an alternative implementation has to be chosen, which made the interface necessary in the first place, and it is surprisingly wordy for that common and simple scenario.

@Named("controllerHooksProvider")
@Singleton
public class JeeControllerHooksProvider implements ControllerHookProvider {

    private Collection<ControllerHook> controllerHooks = new HashSet<>();

    @Inject
    public void setControllerHooks(@Any Instance<ControllerHook> hooks) {
        for (ControllerHook hook : hooks) {
            controllerHooks.add(hook);
        } // for
    } // setControllerHooks()

    public Collection<ControllerHook> getControllerHooks() {
        return controllerHooks;
    } // getControllerHooks()

} // JeeControllerHooksProvider

This really didn't invite me to go deep into JEE again. Additionally it is missing any of the configuration file based bean definitions avoiding much of the interfacing of classes where in fact just two injected values make the difference. With the Springframework configuration files and auto scanning of beans go hand in hand. JEE intentionally left out this part.

Donnerstag, 24. Juli 2014

Fast on the first Request

It seems to be common sense with so many developers, that "expensive" things that only happen once are not that bad. And when those things happen at the wrong point in time that it is a good idea to move them to the "application startup".
So most of us are not very surprised and consider it no problem, if this startup takes some time.

Welcome to the Cloud

At a first sight, deploying my applications in the cloud is exactly the same as it used to be. I follow accepted standards, use common frameworks and respect many best practices. So any runtime platform supporting this should do the job. Cloud in that case means, that many more aspects of deployment and operations get automated - including that starting and stopping of additional instances, load balancing and so on.
But wait... When does the platform get the idea to start new instances?

Let the Customer wait?

It's just load. And load means, that Customers issued HTTP Requests and are awaiting responses in time. From research we know that in time means something around 3s. But when a user request leads to the start of a new instance to handle the load it's quite too late to do all the expensive stuff. Application startup is not the right point in time anymore.
Additionally you can learn, that all those nice precomputations simply re-calculate values the instances from yesterday or the other instance on the other machine already learned. And of course it still is a good idea to have all those values at hand - meaning to "cache" them. But very many of the values which are not coded or configured into the application deployment stay the same as long as the deployment environment doesn't change. Or they stay the same unless the application itself changes them and thus is able to tell when really to re-calculate derived values in the caches.

Examples for all this are:
  • Database derived value caches like query results with calculations based on the result, where just this application writes to the database. Those values might be needed at startup but stay constant until the application itself changes the database contents.
  • Classpath scanning to auto-discover software components. The developers and deployers wont want to collect the lists manually or at build time but the components don't change after the deployment. And definitely not on every startup.
  • Webtemplates and other codes which need to be fetched and prepared for use (e.g. get compiled). Those codes get prepared on every startup of the application while they change not that frequently and especially not on application startup. Obviously they add a big amount to the the response time since those codes need to be executed for the generation of the response.

Just use a Cache

What? Oh, well not that Cache. This Cache. This cache doesn't need to be as fast as the in memory caches already available and I didn't want to re-invent the wheel. They just need to have some values at hand some other instance already calculated and they have to be changed when these values change - which is - as already pointed out - not as frequent as other runtime values.
The values in the cache still are volatile and can be re-calculated by the application at any time. But they should be persisted for some time to be available at startup and reduce the time for the first request in web applications.
The jsr107 cache implementation in the Google App Engine is an example of exactly this approach. I wrapped this cache as a PersistentRestartCache in Tangram and cut reponse times for the first request to a third (or half as the worst case scenario) from around 30s to 11s. For all other platforms available for the Tangram dynamic webapplication framework I presented a simple (maybe too simple) file based implementation which does most of the job as well.
Still this leaves out the classpath scanning of the Springframework or dinistiq. In my environment the Springframework uses between 3s and 7s and dinistiq around 4s to do the basic application setup based on classpath scanning and additional configuration files. So half of the time the startup deals with the framework code and not the applicaiton startup itself. And this value only was achieved by nearly brutal reduction of the portion of the classpath to be scanned creating a "components" package where all autoscanned components reside for the Springframework and dinistiq.

Resume

Caches are a good Idea. Applications with a dynamic deployment are a good idea. Thinking of the application startup as a point in time where long calculations might take place is not that much of a good idea (at least anymore) where instances should be automatically braught up depending on load and not human decisions.
So simply bring together caches with persistence and knowledge when those caches can be invalidated. As an example I did this for the tangram web application framework and had quite some success on the Google App Engine, run@cloudbees, and OpenShift cloud platforms for the Java world.
The last but also important point: Optimizing the application startup in general is worth the time nowadays.

Donnerstag, 17. Juli 2014

Not that Groovy

Ist es eigentlich schon allgemein akzeptiert, daß Web Anwendungen nicht mehr davon ausgehen dürfen, daß es für sie eine entspannte Startphase gibt? Daß sie auch beim ersten Request für den Endbenutzer ausreichend schnell antworten sollten? Es klingt immer noch häufig so, als wären "teure" Dinge, die nur einmal in der Anwendung passieren kein so großes Problem und das man dieses "teure" einfach auf die Startphase verschiebt.
Alles was ich hier schreibe, ist nur für Projekte relevant, in denen ein Deployment nennenswerte Kosten verursacht oder es aus anderen Gründen keine gute Idee ist, ständig die gesamte Software installieren. Mir war diese Stabilität mit großen Systemen auf Basis CoreMedia immer ein so großer Vorteil, daß ich das auch mit Tangram im kleinen nachgezeichnet habe (bzw. dort sogar für die großen Projekte vorgezeichnet habe). Dennoch muß man auf wechselnde Anforderungen genauso dynamisch reagieren, wie auf die Anfragen der Kunden an das System.

Messungen mit Groovy-Code in der Datenbank

Um die Software einfacher und schneller an Anforderungen anzupassen, habe ich mich ja vor Jahren auf Groovy-Code in einem irgendwie gearteten Repository festgelegt. Wir kennen das von Stylesheets, JavaScripts, Templates (z.B. Freemarker, Apache Velocity).
Damals habe ich für einen großen deutschen Telekommunikationskonzern eine Reihe von Messungen durchgeführt, die deutlich belegt haben, daß der Zugriff auf eine fertig kompilierte Groovy-Klasse - oder gar eine vorher abgelegte Instanz davon - keine Nachteile gegenüber dem Java-Code des statischen Programmteiles hat.
Für den einzelnen Request einer Webanwendung ist es also vollkommen egal, ob der Code nun deployed wurde oder in der Datenbank liegt.
In der Folge hat die typische Tangram Webanwendung kaum noch Anteile an Java-Code (siehe z.B. amor oder dragon-map usw). Es wird alles über Templates mit Apache Velocity für die Darstellungsschicht und Groovy-Code für die Geschäftslogik oberhalb der Datespeicherung bis hin zur URL-Struktur erledigt. (Mit JDO kann sogar ein Teil des Datenmodells so in den dynamisch veränderbaren Teil verlagert werden.)
Die Effizienz bei der Ausführung wird durch einfaches Kompilieren und Instanziieren beim Speichern der Codes erreicht.
Und hier genau schlummert ein Problem, das sich mir nun richtig stellt, wo einiges an "dynamischen" Codes (also solchen in der Datenbank, die potenziell häufiger geändert werden) zusammengekommen ist:
Im Gegensatz zu den Aufrufen der Anwendung und dem direkten Benutzen der Instanzen erhalten wir beim Start des Systems einen deutlichen Performance-Nachteil, da hier nun einiges an Datenbankabfragen und Kompiliervorgängen sowie Instanziierungen zusammenkommt. In der Summe ist das deutlich aufwendiger als die entsprechenden Vorgänge mit statischem Java-Code.
Warum interessiert der Start aber nun gerade wieder? In den meisten Szenarien kommt so etwas doch nicht gerade häufig vor.

Willkommen in der Wolke

Für einen selbstbetreuten "on premise" Servlet/JSP Container mag das stimmen, aber gerade für die "billigen" Modelle in der Wolke und z.B. auch die Google App Engine, OpenShift oder run@cloudbees stimmt das überhaupt nicht, da hier zum Skalieren und bei Nichtbenutzung der Trick gerade darin besteht, Instanzen der Anwendungen herunterzufahren oder neue hinzuzustellen.
Bei Instanzen auf der Google App Engine, bei denen keine sonstigen Zusicherungen eingestellt sind, führt das zu einem netten Hinauf- und Herunterfahren wie ein Jojo. Bis Version 0.7 beinhalten alle Tangram Anwendungen einen Cron-Job, der genau das verhindern sollte, indem die Anwendung sich selbst aufgerufen hat. Vor einiger Zeit habe ich mich dem Problem nun endlich gestellt und versucht, die Startup-Sequenz einmal zu tunen. In der Ausgangslage sind wir bei einem Erst-Start im Bereich von gerne 30s abgelaufene Zeit.

Spring Tuning

Das ist nicht das erste Mal: Bereits früher hat das Springframework gezeigt, daß es den Komfort für den Entwickler mit "Kosten" beim Start belegt. Der Scan der Klassen nach Annotationen und Komponenten sowie deren vollautomatisches Zusammensetzen dauert auf einem mittelprächtigen Rechner zwei bis drei Sekunden, in der App-Engine jedoch gerne bis zu acht Sekunden, die ein Kunde am Browser warten müßte. Das konnte damals durch das Eingrenzen beim Scan reduziert werden:

    <context:component-scan base-package="org.tangram" />

Die Angabe des Base-Package ist hier wichtig und grenzt den Bereich der Klassen, die geprüft werden, nachhaltig ein, was einige Sekunden bringt. In Zukünftigen Versionen sollten hier noch weitere Einschränkungen greifen, weil das "base package" org.tangram immer größer wird und über eine Anzahl von Jars verteilt ist. Ab Version 0.8 müssen daher all Komponenten, die gerne gescannt werden möchten "org.tangram.components" mit Vornamen - also Package-Namen - heißen.

    <context:component-scan base-package="org.tangram.components" />

Dazu wurden einige Klassen in den Paketen verschoben und es bringt wieder einmal ein paar Bruchteile von Sekunden ein. Lohnend, aber es durchbricht die fachlich Sortierung der Klassen ein wenig. (Aus org.tangram.jdo wird org.tangram.components.jdo - man findet sich also schon zurecht)

Caching

Moment! Wieso Caching? Das Füllen der Caches - soweit zu dem Zeitpunkt hilfreich oder notwendig  - ist doch gerade das, was den Start u.a. so langsam macht. Aber die Inhalte der Caches werden dann bei Start genau dieselben sein, wie beim letzten Herunterfahren - oder dieselben, die schon eine andere Instanz berechnet hat. Es wäre also ganz nett, wenn man einfach auf diese Wert zurückgreifen könnte.
Das tut Tangram letztlich auch bereits seit langer Zeit in der Google App Engine: Der Scan nach Klassen, die zur Persistenz-Schicht gehören, fällt in die gleiche Rubrik wie der Scan des Spring-Framework und die Daten, die sich die BeanFactory hier merken muß, werden in der Google App Engine im Memory Cache über die JSR107 Schnittstelle gespeichert.
Diese Schnittstelle soll uns nun helfen, noch mehr Vorgänge nur so häufig auszuführen, wie es notwendig ist, unabhängig davon, ob es wir von einem Start-Request sprechen oder mitten in der Arbeit sind. Das macht die Webanwendungen Wolken-tauglicher als sie es bisher sind.
Die Benutzung dieses Caches ist denkbar einfach:

try {
  CacheFactory cacheFactory = CacheManager.getInstance().getCacheFactory();
  jsrCache = cacheFactory.createCache(Collections.emptyMap());
} catch (CacheException ce) {
  log.error("()", ce);
} // try

Und danach ist es mit put() und get() getan, wobei man natürlich immer gewahr sein muß, daß einmal nichts im Cache ist. - Und einige kompliziertere Sache - obwohl java.lang.Serializable markiert - kann man zwar hineinstopfen, findet sie in der Admin-Oberfläche, bekommt sie aber nicht mehr heraus.
Das gilt leider insbesondere für per JDO persistierbare Objekte und kompilierte Java Klassen.

Tunen wo es wehtut

Langsam ist bei einem leeren Cache das beziehen aller Codes, die für die Website notwendig sind, aus dem Datastore. Hier geht es schon bei wenigen Codes eher um Sekunden als die oben zitierten Bruchteile. Also enthält Tangram ab Version 0.8 einen Simplen, persistenten Query-Cache, der genau wie der transiente Cache die IDs der Objekte der Ergebnislisten den Queries erfaßt und ablegt. Dieser Cache bringt der gesamten Anwendung bei ersten Aufrufen einer Instanz eine deutlichen Schub bei abfragebasierten Anwendungen.
Mit diesen ID-Listen muß man sich leider immer noch an den Datastore wenden, um die Objekte zu beziehen, da diese Objekte ja nicht in den Cache gelegt werden konnten.
Diesen Schritt aber wiederum kann man sich bei den Codes gut sparen, in dem man sie beim Lesen in reine transiente Objekte umwandelt (hey, es sind Codes!) und sie dann im persistenten Startup Cache ablegegen zu können.
Jetzt ist eine der längsten Phasen beim Starten das Kompilieren der Groovy-Klassen. Hier den Binärcode zu cachen hat sich bisher als nicht umsetzbarer Ansatz erwiesen. Aber evtl. kommt das ja auch noch. So ist der limitierende Faktor beim Startup auf die Anzahl der Groovy-Codes und den Compiler gesetzt. Mehr ging bisher nicht, aber es hat den Startup in der Zeit gedrittelt. - Und die Ansätze daraus sind verallgemeinerbar für Anwendungen in der Wolke.

Dienstag, 15. Juli 2014

Google App Annoyance - aka Engine

Together with the missing support for the Servlet 3.0 specification in Google App Engine (and, yes, there are still too many situations were this specification level ist not available) we are reading from Google for their App Engine that classpath scanning is an issue on that platform and that this is one of the reasons that kept them from supporting this version of the servlet specification.
What I didn't read from Google is, that they noticed that the Servlet 3.0 version is (arguably) one of the most important steps since Java came to the web. There is no major framework in the Java world anymore not doing any classpath scanning right now. The mentioned Springframework I'm using is just one example.
After some five years of work with the Google App Engine and Java as the only language in use, I changed my mind and don't consider the App Engine as one of the first choices in cloud platforms for the Java world anymore.
And this is really just because of these two small issues.
Like very many others I'm using the Springframework - with component scan and thus with classpath scanning. Any first request to an instance take ages. But "first requests" are common in the cloud, where instances need to be shut down and brought up depending on load. And the cloud is what Google App Engine is about, isn't it?
I invested quite some effort to learn how to be fast on the first request while still using the Springframework and even developed my own stripped down, minimalistic Dependency Injection environment for the application setup (dinistiq) to only have the features I'm using at hand. But this all didn't help to make the end user experience satisfying. Things feel slow.
So in the end this all gave the push for Tangram to support that many new platforms, use the CI Features and Repositories at cloudbees, enjoy the command line access of OpenShift. This brought options of different frameworks and I learned much about cloud deployment and operation scenarios. So in that respect we should be thankful for the Google App Engine weakness.
But it still is the source of some level of complexity and number of artifacts flying around in what I call my dynamic web application framework Tangram. Also this currently renders Google App Engine the second best cloud platform for Java while still having great web based monitoring tools.

Montag, 14. Juli 2014

Ein weiterer Tangram Nutzer

Die Ponton GmbH aus Hamburg hat nun seit einiger Zeit auch eine Webanwendung mit Tangram im produktiven Betrieb. Dabei habe ich sie natürlich selbst in diese Richtung geschubst, um schnell Ergebnisse vorweisen zu können, aber weder war ich der einzige Entwickler noch hab es ausreichende Gegenwehr.
Auf dieser Basis wird anscheinend das Projekt auch mit neuen Anforderungen weiterentwickelt.
Als eher konservatives Layout wird hier noch das Springframework genutzt und auf JPA als Persistenzschicht gesetzt. URL-Formate als Groovy-Codes in der Datenbank und dort auch jede Menge Busineslogik waren aber gerade der Gewinner bei der laufenden Anpassung von Kleinigkeiten. Nun ist nicht mehr jedesmal ein Deployment erforderlich wie bei der Vorgängerlösung (in PHP).
Aber insbesondere danke, daß ich das laut sagen darf.