Finishing the ScriptEditor (re #294)

- Moving lazy variable initialization to a CatalogDB initializer
- Cleanup and document Vizier init sequence.
- Notify user on schema migrations
- Automatically drop documentation modules from scripts.
- CSS polish for ScriptEditor
- Polishing ScriptEditor interface
- Feedback on save
- Save automatically reassigns the URL from project/branch -> script if necessary.
- Added a widget to put all of the URL futzing code in one place.

Author: okennedy

vizier/backend/src/info/vizierdb/Vizier.scala

@@ -91,19 +91,6 @@ object Vizier
         // Read the above comment before modifying connectionTimeoutMillis please.
       )
     )
-
-    // The following initialize ScalikeJDBC's lazy variables
-    catalog.Project.columns;
-    catalog.Branch.columns;
-    catalog.Workflow.columns;
-    catalog.Cell.columns;
-    catalog.Branch.columns;
-    catalog.Artifact.columns;
-    catalog.InputArtifactRef.columns;
-    catalog.OutputArtifactRef.columns;
-    catalog.Result.columns;
-    catalog.Script.columns;
-    catalog.ScriptRevision.columns;
   }
 
   def initSpark() =
@@ -193,20 +180,23 @@ object Vizier
       return
     }
 
+    // Set up the working directory in environments and properties
+    setWorkingDirectory()
+
     // Check for non-mandatory dependencies
     println("Checking for dependencies...")
     PythonProcess.checkPython()
 
     // Set up the Vizier directory and database
     println("Setting up project library...")
     if(!config.basePath().exists) { config.basePath().mkdir() }
-    initSQLite()
-    Schema.initialize()
+    initSQLite()                // Connect to the DB
+    Schema.initialize()         // Init the DB and/or apply schema migrations
+    CatalogDB.initialize()      // Load ScalikeJDBC state
+    bringDatabaseToSaneState()  // Clean up 'running' transactions (e.g., from a system crash)
     // initORMLogging("warn")
-    bringDatabaseToSaneState()
-    setWorkingDirectory()
 
-    // Set up Mimir
+    // Set up Spark/Mimir/etc...
     println("Starting Spark...")
     initSpark()
 

vizier/backend/src/info/vizierdb/catalog/CatalogDB.scala

@@ -58,4 +58,30 @@ object CatalogDB
 
   def withDBReadOnly[T]( op: DBSession => T ): T = 
     traceLongHolds( DB.readOnly { implicit s => op(s) })
+
+  /**
+   * Initialize ScalikeJDBC's lazy variables
+   * 
+   * ScalikeJDBC involves a number of lazy variables that store the actual column names.  They are
+   * initialized by setting up a database connection.  Because SQLite is not reentrant, it's not 
+   * possible to initialize them from within a DB Session.  Instead, you get a DB 'timeout' as the 
+   * initializer tries and fails to acquire a connection.
+   * 
+   * Instead of hoping that they get initialized outside of a session, this method explicitly forces 
+   * materialization of all of the lazy variables
+   */
+  def initialize(): Unit =
+  {
+    Project.columns;
+    Branch.columns;
+    Workflow.columns;
+    Cell.columns;
+    Branch.columns;
+    Artifact.columns;
+    InputArtifactRef.columns;
+    OutputArtifactRef.columns;
+    Result.columns;
+    Script.columns;
+    ScriptRevision.columns;
+  }
 }

vizier/backend/src/info/vizierdb/catalog/Schema.scala

@@ -76,9 +76,11 @@ object Schema
     val requiredMigrations = MIGRATIONS.drop(currentVersion)
     if(requiredMigrations.isEmpty){ return }
 
+    println(s"... updating vizier.db (old version: $currentVersion; new version: ${MIGRATIONS.size})")
+
     CatalogDB.withDB { implicit session => 
-      for((migration, idx) <- requiredMigrations.zipWithIndex){
-        logger.info(s"Applying Migration ${idx + currentVersion}")
+      for((migration, idx) <- requiredMigrations.zipWithIndex){ 
+        logger.info(s"Applying Migration ${idx + currentVersion}: ${migration.getClass.getSimpleName.replace("Migration", "")}")
         logger.trace(migration.sql)
         migration.apply
       }

vizier/backend/test/src/info/vizierdb/test/SharedTestResources.scala

@@ -22,6 +22,7 @@ import java.nio.file.{ Files, Paths }
 import scala.sys.process.Process
 import info.vizierdb.commands.mimir.geocoder.Geocode
 import info.vizierdb.commands.mimir.geocoder.TestCaseGeocoder
+import info.vizierdb.catalog.CatalogDB
 
 object SharedTestResources
 {
@@ -66,6 +67,7 @@ object SharedTestResources
         // Reset the database
         Schema.drop
         Schema.initialize
+        CatalogDB.initialize()
 
         // And initialize testing
         DummyCommands.init

vizier/shared/src/info/vizierdb/serialized/VizierScript.scala

@@ -42,7 +42,18 @@ object VizierScript
       branchId = branchId,
       workflowId = workflow.id,
       modules = workflow.modules.map { module =>
-        VizierScriptModule.Inline(module, enabled = {module.statev2 == types.ExecutionState.DONE})
+        val moduleRanSuccessfully = {module.statev2 == types.ExecutionState.DONE}
+        val moduleInvolvesDocumentation = 
+          module.command.packageId match {
+            case "docs" => true
+            case _ => false
+          }
+        val shouldEnable = (
+          moduleRanSuccessfully 
+            && !moduleInvolvesDocumentation
+        )
+
+        VizierScriptModule.Inline(module, enabled = shouldEnable)
       }
     )
   }