[Subversion] / PEAK / TODO.txt  

Diff of /PEAK/TODO.txt

Parent Directory | Revision Log

version 959, Thu Apr 3 14:33:53 2003 UTC version 1659, Fri Jan 30 00:28:02 2004 UTC
Line 1 
Line 1 
 Open Issues/To-Do Items  Open Issues/To-Do Items
   
  Targeted for 0.5 Alpha 1   Targeted for 0.5 Alpha 3
   
     * Fix issue w/reading XMI 1.1 files where metamodel has nested packages      * peak.events migration
   
     * Clean up __class_provides__ so that it isn't just something randomly        - 'running.IProcessProxy' should use conditions/values instead of
       poked in by binding.Once(); it needs a clean infrastructure/interface.          listeners, and related command/supervisor tools should also be ported
   
     * Refactor "Pythonic" UML 1.3 extensions so they're usable for most        - Add tests for Twisted support
       MOF-aligned metamodels (e.g. all UML and CWM versions)  
   
     * Remove deprecated items from peak.model: Feature, Reference, Package,      * peak.ddt
       Model, Namespace, DataType, etc.  
   
     * Get XMI writing in place, w/transaction support        - A bit more doc and demos
   
     * Updated reference docstrings for 'peak.api', 'peak.binding',  
       'peak.config', 'peak.exceptions', and 'peak.naming'.  
   
     * Finish tutorial chapter 2 (?)  
   
  Targeted for 0.5 Alpha 2  
   
     * On-the-fly class combination (think "runtime module inheritance",  
       without the modules) for DMs.  
   
     * ZConfig support, probably in the form of generating ZConfig schemas  
       from 'peak.model' or MOF models, but maybe in the form of 'fromZConfig'  
       constructor methods as well.  
   
     * 'peak.running' refactorings: use standard 'logging' module's log levels,  
        add 'logging' distro to 'peak.util' for 2.2 backward compatibility,  
        make daemons based on 'peak.model' (or at least define ZConfig schemas),  
        and possibly adjust cluster stuff to work off ZConfig primary and  
        generate clustertools file(s).  App startup tools based on ZConfig and  
        PEAK-style .ini files.  
   
     * 'peak.naming' refactorings: 'peak.model'-based syntax utilities for  
        creating address syntaxes.  
   
  Targeted for 0.5 Final Release (or sooner)  
   
     General  
   
         * Tutorial complete through chapter 4  
   
     peak.storage  
   
         - unit tests for more complex object scenarios: references, thunks..?  
   
         - lock management interfaces/API  
   
         - docstrings for reference  
   
     peak.model  
   
         - clean up TW docstrings & interfaces  
   
     peak.naming  
   
         - useful example "flat" naming context (e.g. like AppUtils.URLkeys)  
   
         - useful example hierarchical naming context (e.g. like JNDI's LDAP  
           context or filesystem context)  
   
         - rework smtp: to return a factory object that supports open().  
           Also think about whether smtp should move elsewhere. Maybe  
           there should be peak.network or peak.internet for things like  
           smtp, ftp, etc contexts?  
   
     peak.running  
   
         - make 'cluster' parser complain about things that would cause  
           the clusterit tools to choke or barf on the file, or which would  
           produce ambiguous or unintended results.  
   
         - simple daemons comparable to those in MetaDaemon, unit tests  
   
         - docstrings for reference  
   
   
     peak.config  
   
         - "Rule"-oriented configuration files (section specifies component  
           rather than property name prefix), so that daemons and other simple  
           apps can be fully configured and run via a config file.  
   
     peak.util  
   
         - docstrings for reference  
   
         - more unit tests?  
   
   
   
   
    Targeted for 0.5 Alpha 4
   
       * 'ILoadingService', 'IObjectLoader' and 'load:kind@URL' scheme
   
       * peak.config
   
         - "service zones" or "service groups" to do smarter service groupings
   
         - Replace IMainLoop activity monitoring with a plugin-based mechanism?
   
         - plugin keys ordered by definition sequence, rather than randomly
   
         - ZConfig factory support to allow "smart" interpretation of strings,
           section names, etc.
   
         - Writable and subscribable config sources, including editable .ini's
   
       * peak.running.logs
   
         - Separate formatters from publishers
   
         - ZConfig schema for logging plugins, allowing multiple handlers/category
   
       * peak.running.commands
   
          - Add "error formatting" and "error reporting" services
   
          - Add option parsing framework based on optparse (backported to 2.2)
   
       * peak.binding
   
         - Cleanup/document attribute binding interface
   
       * peak.running.timers
   
         - Factor out state into separate object so timers aren't shared state
   
         - Look at possible integration of peak.query and cursor formatters, in
           order to view stats as an in-memory mini-DB w/reporting.
   
   
       * Transaction/storage refactorings
   
         - transaction scopes for commands and tasks
   
         - integrate locks with transactions
   
         - transactable persistent queues
   
         - DM.find()/DM.get() queries
   
       * peak.web
   
  Future Releases         - clean up DOMlet parse/build framework (e.g. add line number info)
   
   (Note: some of the below is held-over from TransWarp and may no longer be         - default error templates, w/useful info
   relevant as written, they are being kept on this list as placeholders for  
   ideas or problem areas that may need to be re-considered in future.)  
   
          - A set of simple, basic form controls that handle value rendering only
            (form metadata, validation, etc. will be handled by peak.web.forms in
            a later release)
   
   Simulator/Module Inheritance         - try/catch DOMlet (and related error rendering interface/framework)
   
     * Need a strategy for handling "del" operations; they are currently         - allow use of // in DOMlets' data paths, to access resource space
       untrapped.  This might be okay under most circumstances, but need to  
       consider edge cases.  
   
     * 'makeClass()' should probably become part of the core API, where         - DOMlets for layout/region, as defined in "A layout framework":http://www.eby-sarna.com/pipermail/transwarp/2003-August/000684.html
       it can be used to resolve __metaclass__ conflicts during the first  
       pass of importing a module (prior to running 'setupModule()')  
   
          - conditional GET support (last modified/ETag) for static resources
   
   Messaging/ObjectSpaces         - image resources
   
     * Support for sending and receiving remote cache invalidation         - the return of the Specialist
       messages between RecordManagers.  
   
       * Have a way to easily in-line custom component usage (e.g. automatically
         create a subclass component with specified 'Obtain' or 'Make' bindings to
         get its configuration).
   
       * Drop 'persistence' package, since ZODB 4 has been derailed.  Change to
         "state-delegation" model, which will integrate better with 'peak.query'.
   
   
   
    Targeted for 0.5 Beta 1
   
       * Up-to-date reference docstrings for all packages
   
       * Finish tutorial chapter 2 (?)
   
       * Web publishing framework sufficient to deploy page-based or object-
         published apps
   
   
   
    Targeted for 0.5 Final Release (or sooner)
   
       peak.storage
   
           - unit tests for more complex object scenarios: references, thunks..?
   
       peak.naming
   
           - useful example "flat" naming context (e.g. like AppUtils.URLkeys)
   
           - useful example hierarchical naming context (e.g. like JNDI's LDAP
             context or a filesystem context)
   
           - rework smtp: to return a factory object that supports open().
             Also think about whether smtp should move to peak.net?
   
   
   peak.model  
   
     * Implement WarpCORE-oriented structural model, w/Querying support  
   
     * "Indexed" version of in-memory model?  
   
   
     Queries  
   
        * Refactor to use interfaces, if appropriate  
   
        * Incorporate into AbstractModel?  
   
          - Pros:  
   
            * Queries always available  
   
            * Each StructuralModel implementation can easily include its own  
              performance-tuned version of the basic items.  
   
          - Con: default implementation doesn't perform well on large datasets  
   
        * How much of framework needs extensibility?  Should the predicate  
          classes be placed in the StructuralModel's namespace so that predicates  
          have their meaning assigned by the StructuralModel implementation?  
   
   
    Targeted for version 0.6
   
      * Functional/acceptance tests that access "real" databases, LDAP, etc.
   
      * Get XMI writing in place, w/transaction support
   
      * Generate UML 1.5 and CWM 1.0 and 1.1, and add them to the
        'setup.py' package lists.
   
      * On-the-fly class combination (think "runtime module inheritance",
        but possibly without the modules) for DMs.
   
      * Lock management interfaces/API for peak.storage
   
      * Support for sending and receiving remote cache invalidation
        messages between DataManagers.
   
      * Make 'cluster' parser complain about things that would cause
        the clusterit tools to choke or barf on the file, or which would
        produce ambiguous or unintended results.  (Or replace with ZConfig
        schema that can generate clusterit files.  And/or replace clusterit
        tools with PEAK ones.)
   
   
   
Line 203 
Line 191 
   
   
   
   peak.metamodels.uml  
   
     * Helper methods in Elements & Services for marshalling, common queries, etc.  
   
     * Generator framework  
   
       - Tagged values in stereotypes vs. main values?  
   
       - Should tagged values be copied directly into templates?  Treated as  
         Python expressions?  
   
       - Should Services be generated using an Element class' "static"  
         (class-scope) methods/attributes?  
   
         - Are association-ends scoped?  
   
         - Would it be better to seperate them?  
   
       - What determines whether an implemented Service actually stores objects  
         or delegates this to its subclass services?  
   
     * Simple Zope product demo (upload XMI, then browse the model via the web)  
   
   
   


Generate output suitable for use with a patch program
Legend:
Removed from v.959  
changed lines
  Added in v.1659

cvs-admin@eby-sarna.com

Powered by ViewCVS 1.0-dev

ViewCVS and CVS Help