[rkward-cvs] SF.net SVN: rkward: [2129] branches/KDE4_port

[tfry at users.sourceforge.net]

Revision: 2129
          http://rkward.svn.sourceforge.net/rkward/?rev=2129&view=rev
Author:   tfry
Date:     2007-10-25 10:46:23 -0700 (Thu, 25 Oct 2007)

Log Message:
-----------
Merge 2126 from trunk

Modified Paths:
--------------
    branches/KDE4_port/ChangeLog
    branches/KDE4_port/TODO
    branches/KDE4_port/rkward/rbackend/rpackages/rkward/DESCRIPTION
    branches/KDE4_port/rkward/rbackend/rpackages/rkward/R/internal.R

Modified: branches/KDE4_port/ChangeLog
===================================================================
--- branches/KDE4_port/ChangeLog	2007-10-25 17:39:00 UTC (rev 2128)
+++ branches/KDE4_port/ChangeLog	2007-10-25 17:46:23 UTC (rev 2129)
@@ -1,4 +1,3 @@
-- Analysis->Moments->Moment plugin did not use the order-parameter	TODO: backport rev. 2122
 - code completion in script editor shows icons for the different object types
 - settings dialog is now shown in "page list" mode
 - port to KDE 4			(TODO)
@@ -6,6 +5,8 @@
 ---- KDE 3 versions
 
 
+- Fixed: Analysis->Moments->Moment plugin did not use the order-parameter
+- Fixed: Crash on simple assignments in globalenv() with R < 2.4.0
 - Fixed: Meta information (labels) did not get shown correctly in the data editor
 - Add command line option "--debugger" for debugging purposes
 - Fixed: Pressing the add button in a varslot with no object selected could crash RKWard

Modified: branches/KDE4_port/TODO
===================================================================
--- branches/KDE4_port/TODO	2007-10-25 17:39:00 UTC (rev 2128)
+++ branches/KDE4_port/TODO	2007-10-25 17:46:23 UTC (rev 2129)
@@ -141,6 +141,13 @@
 			- likely, package environments should try to fetch vignettes, when "updating"
 
 Internal stuff:
+	- Workspace browser / code completion
+		- we might make this a bit less pro-active:
+			- only objects directly on the search path need to be fetched at once
+			- children of these objects do not need to be investigated until we first try to access a child of the object.
+			- i.e. RContainerobject::children() should handle fetching the child structure (numChildren() should probably be populated pro-actively).
+			- isParentOf() should not be affected, as this can only be true for a child object that has previously been investigated
+			- need to make sure, completions are not blocking
 	- Handling fonts:
 		- http://sourceforge.net/mailarchive/forum.php?thread_id=31631211&forum_id=12970
 	- Look at package odfWeave, and find out, how to use this

Modified: branches/KDE4_port/rkward/rbackend/rpackages/rkward/DESCRIPTION
===================================================================
--- branches/KDE4_port/rkward/rbackend/rpackages/rkward/DESCRIPTION	2007-10-25 17:39:00 UTC (rev 2128)
+++ branches/KDE4_port/rkward/rbackend/rpackages/rkward/DESCRIPTION	2007-10-25 17:46:23 UTC (rev 2129)
@@ -1,6 +1,6 @@
 Package: rkward
 Title: Provides some helper functions for the RKWard frontend
-Version: 0.4.8
+Version: 0.4.8.1
 Author: Thomas Friedrichsmeier and the RKWard Team
 Description: Most of the functions in here are really only needed for the internal communication between RKWard and R. There are also a few functions, which allow access to RKWard or RKWard specifics. Most is not implemented, yet.
 Maintainer: RKWard-devel mailing list <rkward-devel at lists.sourceforge.net>

Modified: branches/KDE4_port/rkward/rbackend/rpackages/rkward/R/internal.R
===================================================================
--- branches/KDE4_port/rkward/rbackend/rpackages/rkward/R/internal.R	2007-10-25 17:39:00 UTC (rev 2128)
+++ branches/KDE4_port/rkward/rbackend/rpackages/rkward/R/internal.R	2007-10-25 17:46:23 UTC (rev 2129)
@@ -167,26 +167,47 @@
 	}
 }
 
-".rk.make.watch.f" <- function (k) {
-	# we need to make sure, the functions we use are *not* looked up as symbols in .GlobalEnv.
-	# else, for instance, if the user names a symbol "missing", and we try to resolve it in the
-	# wrapper function below, evaluation would recurse to look up "missing" in the .GlobalEnv
-	# due to the call to "if (!missing(value))".
-	get <- base::get
-	missing <- base::missing
-	assign <- base::assign
-	.rk.do.call <- rkward::.rk.do.call
-	invisible <- base::invisible
-
-	function (value) {
-		if (!missing (value)) {
-			assign (k, value, envir=.rk.watched.symbols)
-			.rk.do.call ("ws", k);
-			invisible (value)
-		} else {
-			get (k, envir=.rk.watched.symbols)
+if (compareVersion (paste (R.version$major, R.version$minor, sep="."), "2.4.0") >= 0) {
+	".rk.make.watch.f" <- function (k) {
+		# we need to make sure, the functions we use are *not* looked up as symbols in .GlobalEnv.
+		# else, for instance, if the user names a symbol "missing", and we try to resolve it in the
+		# wrapper function below, evaluation would recurse to look up "missing" in the .GlobalEnv
+		# due to the call to "if (!missing(value))".
+		get <- base::get
+		missing <- base::missing
+		assign <- base::assign
+		.rk.do.call <- rkward::.rk.do.call
+		invisible <- base::invisible
+	
+		function (value) {
+			if (!missing (value)) {
+				assign (k, value, envir=.rk.watched.symbols)
+				.rk.do.call ("ws", k);
+				invisible (value)
+			} else {
+				get (k, envir=.rk.watched.symbols)
+			}
 		}
 	}
+} else {
+	# see above, but '::' operator was more picky in R < 2.4
+	".rk.make.watch.f" <- function (k) {
+		get <- base::get
+		missing <- base::missing
+		assign <- base::assign
+		.rk.do.call <- get (".rk.do.call", envir=as.environment ("package:rkward"))
+		invisible <- base::invisible
+	
+		function (value) {
+			if (!missing (value)) {
+				assign (k, value, envir=.rk.watched.symbols)
+				.rk.do.call ("ws", k);
+				invisible (value)
+			} else {
+				get (k, envir=.rk.watched.symbols)
+			}
+		}
+	}
 }
 
 ".rk.watch.symbol" <- function (k) {


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.