Dec 06 CTP release

2006-12-30
2013-05-28
  • Brian Nalewajek

    Brian Nalewajek - 2006-12-30

    Hi,
      Thankyou to everyone working on the NORMA team for posting the latest release.  I see that you've promissed some new release notes soon.  Don't think I'm going to wait for that to have a look at the new version.  I just did the SP1 install to VS 2005, so we'll see if that has any effect on the NORMA install.  Guess we can expect a flurry of new posts to start the new year - not a bad way to start off!  BRN..

     
    • Matthew Curland

      Matthew Curland - 2006-12-30

      SourceForge was not letting me add release notes, I tried several times.

      A few things to look for:

      Verbalization: Equality/Subset/Exclusion constraints now verbalize property. We also have generated documentation in the <INSTALLDIR>/Xml/Verbalization/Core directory and an improved snippets schema file.

      Generation: A lot of progress has been made on sproc and trigger generation across most of the databases. We also added MySQL and a PHP entity layer (yes, you still need to generate in a C# or VB project: we need an MSBuild-enabled host, which a solution file is not). The PLiX_Implementation entry in the custom tool dialog now has three options under it, so you can spit something other than the full in-memory implementation. I think a couple of checkins in this area went home for the holidays, so there are a few fixes we don't have in yet.

      Stability: I've spent a lot of time fixing bugs this quarter. The designer should be much more stable. We've wrapped most event handlers so an error in one tool window will not destabilize all of them. Still, if you get an exception dialog, I'd recommend saving and restarting the environment.

      Performance: We've concentrated on working through perf issues with large models. A lot of the slowdowns came from the DSLTools framework. We can now import a Orthogonal-toolbox generated model with 980 roles in ~15 seconds. Most of that is the diagram load. This scenario used to take 3-5 minutes for the import and at least 30 for the initial diagram generation. The new model browser is also much faster (the removed DSLTools-provided browser was incredibly slow with large models). The context window is also much faster.

      Model browser: The model browser is brand new and much more performant than it was. This is just a start: we'll add glyphs, custom categorization and grouping, multiselect, etc as we move forward. We just had to get the old one out of there because of the perf issues.

      Context window: The context window layout is much better, and using the window is much more stable. Especially at generation 1 (which still shows quite a bit), you won't see any slow down as you move your selection around.

      General editing:
      1) Reading editing no longer lets you directly edit the {#} replacement fields. BTW, if you have readings with out-of-order replacement fields ("{1} has {0}" for example) you may have to reenter these with the correct reading order. The numbers here refer to the indices of the roles in the ReadingOrder, not the native role order. We'll get inline editing going as well with the editor, but the current redirect-to-the-reading-editor behavior keeps the underlying strings safe for now.
      2) You can now perform all fact shape orientation actions off the context menu (reverse role order, rotate left/right, etc). You'll also see new reading direction indicators that are much cleaner.
      3) You can now add an ExclusiveOr constraint, which is really two synchronized constraints (an Exclusion and a Disjunctive Mandatory). You also have commands to split and combine these constraints.
      4) Drag-drop of shapes between diagrams in the same model is now supported (I checked this in today, I thought it was important). This is always a copy operation now: shapes are duplicated in the new diagram. Multiselect is supported, but all elements must be able to move individually. Also note that you can drag items out of the context window, which makes it easy to rebuild diagrams using the Model Browser and the context window (great for import scenarios). To add a new window, execute the Window/New Window command, then right click the VS-provided tab on the document and create a new tab group. You can now see multiple windows simultaneously.
      5) The Sample Population editor does a better job of moving to the next edit field on commit.
      6) There are a lot more validation rules wrt/implied and contradictory constraint patterns.

      Miscellaneous stuff (not all in setup):
      1) There is code in the repository for using the CodeSmith .NETTiers templates directly off of a DCIL file (our generated relational representation). This piece is relatively well tested.
      2) We've started work on a PLiX version of the .NETTiers templates, which will let us generate .NETTiers code off of an .ORM file without any CodeSmith involvement.

      Basically, we've been going for stability and performance since the last release. We've added some object model support without any UI yet (join paths, implicit binarization of unary facts). There are several big changes pending (a rework of the datatyping and refmode systems, file format name cleanup), but we didn't want to destabilize at this point.

      We're also working through issues of getting the SVN hosted an Neumont instead of SF so that it is easier to pick up bug fixes and build intermediate drops locally.

      Please let us know of any problems with setup.

      It is still the week before New Year's:)

      -Matt

      PS We've had a couple of problems with the toolbox not working after an uninstall/reinstall. If you hit this, try shutting down VS and deleting any .tbd files under Documents and Settings\YOU\Local Settings\Application Data\Microsoft\VisualStudio\8.0 to eliminate any toolbox caching. We're not sure what is causing this problem (we've seen it twice).

       
    • Brian Nalewajek

      Brian Nalewajek - 2006-12-31

      Hi,
        Thanks Matt, for the preview of new release notes. I really need to spend some time looking through these and the old posts to get a good feel for what was and will be changing.  This time of year, I tend to revert to my "Instructions? we don't need no stinking instructions!" mode. It's pretty clear that your team was focused on preformance and stablity in this release.  I only took the release for a brief spin so far, and only looking at a model created with the last release.  The UI does look more stable, with less random/artial windows openning up or not going away.  I did see that the Object Explorer has been simplified - and I take it from your post, that it was a major overhaul, and will be added to as you move forward.
        I hope to have a chance to kick the tires in the coming week, and will keep an eye on these forums.  Good luck with bringing the project hosting home to Neumont - I'm sure that will facilitate the development process.  Also, good luck with the presentation in Jan. you mentioned - hope we'll see some materials from that posted at ORM.net, or elsewhere.  Happy New Year!  (I'll be headed out to First Night in Hartford later - very nice upclose fireworks display at midnight here.  Unlike most years - due to the mild start to winter here), the sound of the fireworks won't be drowned out by the sound of chattering teeth.)  BRN..

       

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks