{13} All Tickets by Milestone (Full description) (76 matches)

We're trying to design a good way for writing and publishing the documentation of a plugin.

The main goal is to have an homogeneous documentation format for all plugins, that is lightweight for the plugin author. Considered elements:

• Documentation available on the wiki
• Common format:
  • Common header with link to plugin usage documentation
  • Short description
  • API javadoc linked from the documentation
  • jar linked from the documentation
  • Detailed description
  • Requirements / Dependencies
  • Examples
• Examples could be wroten in source file and fetched from svn

Using javadoc seems to be a good solution for the developper point of view, so to have a chance to have wroten documentation :-)

Proposed design:

• Developper wrote javadoc at package level using package-info.java (available in new file wizard in NetBeans), using h3 sectioning
• Add a trac plugin that gets this documentation and integrates in the wiki adding generic information (API javadoc and jar links).

Technical hints:

• http://java.sun.com/j2se/1.4.2/docs/tooldocs/javadoc/taglet/overview.html
• http://docs.codehaus.org/display/GEOT/2.6+Generating+Javadoc

For now, if a task fails because of an exception, the exception isn't propagated to the call stack that have created the thread that manages the failing task.

We should find a way (Progress interface, new listener) for retrieving these execeptions at a higher level.

Note: it isn't annoying in the console as the exception is still printed in the output.

When an application containing a groovy plan is reload, the correct plan definition is not reified, it's the default plan that appears.

It could be nice to display a progress bar when an exploration is launched to :

• know the state of exploration
• get a feedback, that is important to be sure that the softwar is still working

Mercurial is integrated in netbeans, also it seams that git is better at automatically merging files… I think we should give a try to git !

TODOs:

• revision ID in trac

The editor to choose the design method should be more explicit and give access to a documentation about the selected method, and its configuration.

add variable panel : need to verify if the variable name already exist

There is a lot of duplicated code for nodes management : variables, structure, libraries, application, … It would be nice to build an abstraction for layer for our purpose. In the same manner, we should create a factory for properties sheet creation. The better example is the case of metada, that is duplicated and not so simple …

Update existing tutorial for using SimExplorer

The ExplorationTask is responsible to build the attached plan using the defined factors. If no factors are defined, I think that the task should not try to built the plan. Indeed, most of implemented plan can implement test for being robuste to this case, but it will be more logical to assert that we have at least one factor defined when building the plan.

Replace sh scripts and ant tasks with maven based solution.

If you set the step value of the rangeDoubleDomain when you define an experimental factor, the exploration crashes with the following stacktrace:

java.util.NoSuchElementException
        at java.util.AbstractList$Itr.next(AbstractList.java:350)
        at org.simexplorer.methods.plan.CompletePlan.computeValues(CompletePlan.java:45)
        at org.simexplorer.core.processors.AbstractArrayPlan.build(AbstractArrayPlan.java:37)
        at org.simexplorer.core.processors.ExperimentalDesign.build(ExperimentalDesign.java:48)
        at org.simexplorer.core.processors.AbstractExploration.buildDesign(AbstractExploration.java:80)
        at org.simexplorer.methods.processors.ExplorationProcessor.prepare(ExplorationProcessor.java:59)
        at org.simexplorer.core.processors.Processor.run(Processor.java:167)
        at org.simexplorer.core.processors.ProcessorsList.process(ProcessorsList.java:79)
        at org.simexplorer.core.processors.Processor.run(Processor.java:170)
        at org.simexplorer.core.processors.ProcessorsList.run(ProcessorsList.java:278)
        at org.simexplorer.core.processors.IDEModularApplication.run(IDEModularApplication.java:54)
        at org.simexplorer.runtime.SimExplorer.start(SimExplorer.java:107)
        at org.eclipse.equinox.internal.app.EclipseAppHandle.run(EclipseAppHandle.java:193)
        at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.runApplication(EclipseAppLauncher.java:110)
        at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.start(EclipseAppLauncher.java:79)
        at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:382)
        at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:179)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
        at java.lang.reflect.Method.invoke(Method.java:597)
        at org.eclipse.equinox.launcher.Main.invokeFramework(Main.java:549)
        at org.eclipse.equinox.launcher.Main.basicRun(Main.java:504)
        at org.eclipse.equinox.launcher.Main.run(Main.java:1236)
        at org.eclipse.equinox.launcher.Main.main(Main.java:1212)

A no entry sign appeared when launching SimExplorer

It correspond to the following message

java.lang.NullPointerException

at org.netbeans.NetigsoModule$DelegateCL.findResources(NetigsoModule.java:201)

Caused: java.lang.reflect.InvocationTargetException

at sun.reflect.GeneratedMethodAccessor2.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) at java.lang.reflect.Method.invoke(Method.java:597) at org.netbeans.core.startup.NbInstaller.loadLayers(NbInstaller.java:586) at org.netbeans.core.startup.NbInstaller.load(NbInstaller.java:328) at org.netbeans.ModuleManager.enable(ModuleManager.java:929) at org.netbeans.core.startup.ModuleList.installNew(ModuleList.java:315) at org.netbeans.core.startup.ModuleList.trigger(ModuleList.java:251) at org.netbeans.core.startup.ModuleSystem.restore(ModuleSystem.java:283) at org.netbeans.core.startup.Main.getModuleSystem(Main.java:169) at org.netbeans.core.startup.Main.start(Main.java:299) at org.netbeans.core.startup.TopThreadGroup.run(TopThreadGroup.java:111)

[catch] at java.lang.Thread.run(Thread.java:619)

With the edit action the data type for variables dont changed after editing

A saved application cannot be reloaded

The aim is to provide a converter between the IDEModularApplication stored in the IDE and an ExplorationApplication in the IS.

Then the IDE will be enable to use the IS Swing client :

• when open action is called in the IDE, the application explorer of the IS is opened for choosing an application, then data are converted, and the application is opened in the IDE
• when save action is called in the IDE, the application is converted and give to the structure of the IS, and the IS stored the application at the correct place according to the version.
• when manage action is called in the IDE, the IS Swing client window is opened, and the user could export the components on the server

Metadata have been removed from inside tasks. They have to be accessible in a new service that will associate metadata to the concerned task.

For example, the list of jars in the groovyprocessor is actually saved as constant in the information system, but it isn't suitable because if the user modify the list of jars, the application should be considered as changed. Constant should be reserved to true exploration parameters.

A sequence node is named by its nested node (recursively if it is a sequence node). For example, if I create such a structure :

ComplexNode : "myStructure"
  \__ SequenceNode
        \___ ComplexNode : "element"
                 \___ Prototype : "value" : Integer

The corresponding class is generated :

public class CmyStructure implements DataContainer, DataModifier {
  public List<CmyStructureCelement> element = [];

  @Override
  public Object getValue(String name) {
    return this[name];
  }
  @Override
  public void setValue(String name, Object value) {
    this[name] = value;
  }
}

Well. First, the naming of the sequence node is ambiguous : your inner node is singular, so you wan't use a plural form, and give the name "element". But, the attribute is a sequence and a plural form ("elements" in this exemple) would be more appropriate.

Then, we need a convenient way to populate the sequence with new elements, or else, user should populate the list like this :

myStructure.element[i] = new org.simexplorer.user.data.CmyStructureCelement();

With the core of SimExplorer, we can build complex DoE, and particularly nested DoE. It is the case when you want to explore some factors with a plan method and some other factors with an other plan method. But, for a convenient purpose, it will be better solution to present these different methods in the same "exploration loop". I propose to display a list of panels in the configuration of the ExplorationLoop component. The user can add and order the method used for the exploration, and select the factors associated to each method. For example:

Exploration Loop Configuration Panel
____________
Complete plan (a1,a2)
[ Configuration of the method ]
____________
LHS plan (a3,a4)
[ Configuration of the method ]

Use data structure to describe the structures of the all the meta-data

Write specifications gathering ticket

Include these tickets: No results

A Tutorial witch explain how to create a plugins is needed

PB : Between version .02RC6 and 0.2RC7 there are changes in the format of tne groovy processor. This conduce to a change in the format of the xml file. So a preceding experiment cannot be loaded : this xml file has to be changed manually.

For a user point of view this type of incompatibility is unacceptable. We have to provide some utilities to move to the new format!!!

Histogram, scatter plot, … (If dependent of a variable ask for it.)

Provide an object that gives access to random number generation, especially in the plan method and input generation processors.

SSJ should be convenient for this.

For example, for the LHS implementation, we need to have access to several distributions with their cumulative function and inverse.

Actually, all logging messages are printed on the output, this is too verbose. It would be better to be able to log in separate files messages concerning

• the platform
• internal errors of SimExplorer
• user bad data errors

See also this doc

Have one logger by class.

For example:

• exporting the P. Auger Predator Prey equa diff. model as a .jar
• loading the jar in groovy script
• puting input SE variables into input model variables
• invoking it

How to run unit tests? Gathering documentation about creation and good practices.

Some concepts are not fully described in the trac doc. For instance, we don't have any idea of what is an Exploration Task or a data channel.

Goals are:

• show the evolution of SimExplorer
• demonstrate the low(!) overhead of SimExplorer

An interesting solution would to use the infrastruture provided by the Nb platform as described in the two following links :

• How To Use UI Gestures Collector In Your App
• Collecting Data on Users of a NetBeans Platform Application (Part 1)

This infrasture can be use for automating all the aspects of the logging:

• Exception reporting and
• slowness detection
• different module/api usage
• UI usage
• ...

I didn't find any obvious way to write a new entry in the blog of the site. Is there one ?

How to garanty that a workflow is reproductible?

The main issue is that all the plugins used in the mole should insure that they don't depend on external resources.

A example of solution: to sign with GPG certified plugins distributed or available on servers. Then, when the mole is executed, if a plugin non certified is used, a warning is displayed to prevent that the reproducibility is NOT garanty.

Actually factors are defined by only a domain and are then expanded at the begin of each iteration of the exploration.

It could be usefull to firts define a definition domain, and then apply transformation function(s).

Depending if the definition domain is finite or not, this transformation should be applied during the sampling process or just after the begin of the iteration on the current value of the factor.

A solution could be to explicit the sampling process in the workflow and give hooks to apply these transformation functions. In fact this process could be composed as follow :

1. Factor Domain Definition (output: domain and its parameters)
2. If finite domain
   1. First Sampling (output: array of values)
   2. Loop on transformation functions (output: array of values)
3. Iteration on values (output: value)
4. Loop on transformation functions (output: value)

Actually, the IDE main window displays the default window title. It would be more usefull to display something like "SimExplorer ${version} — ${Application name} ${Application version}"

See this Geertjan's blog entry