[forge-issues] [JBoss JIRA] (FORGE-699) create a plugin for WildcardImportResolver

[George Gastaldi (JIRA)]

     [ https://issues.jboss.org/browse/FORGE-699?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

George Gastaldi updated FORGE-699:
----------------------------------

    Fix Version/s: 1.2.2.Final
                       (was: 1.2.1.Final)

    
> create a plugin for WildcardImportResolver
> ------------------------------------------
>
>                 Key: FORGE-699
>                 URL: https://issues.jboss.org/browse/FORGE-699
>             Project: Forge
>          Issue Type: Story
>          Components: Parsers / File Manipulation
>    Affects Versions: 1.1.1.Final
>            Reporter: fiorenzo pizza
>            Assignee: fiorenzo pizza
>             Fix For: 1.2.2.Final
>
>
> general notes
> We need to create some WildcardImportResolver (Responsible for providing additional import resolution functionality for situations where classes have referenced a package or wild-card import)
> summary:
> - when start creation of index			--> new-project
> - when update the index				--> re-open project | add|remove plugins to the project | add|remove maven dependencies | add|remove new class inside forge project
> - where save the index				--> always		
> - how information to save for future search	--> className, jarName (null if the class is in the project), packageName, absoluteClassName
> - how search inside the index			--> (ram hashmap? ram db?)
> - how generate new entry for the index		--> cdi event
> general considerations
> 1) when forge create a new-project we should create some structure to add all classes used referenced by project
> 2) we need to add all classes required from some plugin (persistence|scaffolding..)
> 2) we need to add all classes inside maven dependencies
> 3) we need to add all classes created in the project
> 4) we need remove all deleted classes
> 5) we need to remove all classes inside dependecies removed
> 6) we need to update the index files, when forge project is reopen (the user could modify the project out of forge)
> problems
> 1) with a lot of jars, the indexing process should be slow!!
> 2) how synchronize the access to index
> 3) how we should synchronize of index update?
> requirements
> 1) we need some fast library to extract classes from jar files
> 2) we should use cdi event to start a single jar indexing 
> HOW CREATE INDEX FROM JAR:
> possibile libraries: jandex, tattletale, jarindexer
> FIRST CANDIDATE: JANDEX
> i started to hack jandex code, because jandex is very fast.
> Jandex opens the jar files using java jar utilities: JarFile jar = new JarFile(jarFile);
> and for every class generate a ClassInfo object: ClassInfo info = indexer.index(jar.getInputStream(entry));
> For my test, i removed some methods inside jandex: we don't care about annotations inside classes. Also i removed the generation of file index.
> questions:
> 1) why don't use H2, the Java SQL database, for memorize all informations about classes?
> How memorize for a single class:
> 1) classname, jar name (with version), package name, complete name

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira