Fixed bug [2886436fff] - [.txt delete] deletes before start index - This is option 2: don't change the behavior of the text widget, but document it better.

2 files changed, 14 insertions, 7 deletions

diff --git a/doc/text.n b/doc/text.n
index b11363d..1ccadbb 100644
--- a/doc/text.n
+++ b/doc/text.n
@@ -1216,8 +1216,11 @@ If \fIindex2\fR does not specify a position later in the text
 than \fIindex1\fR then no characters are deleted.
 If \fIindex2\fR is not specified then the single character at
 \fIindex1\fR is deleted.
-It is not allowable to delete characters in a way that would leave
-the text without a newline as the last character.
+Attempts to delete characters in a way that would leave
+the text without a newline as the last character will be tweaked by the
+text widget to avoid this. In particular, attempts to delete complete
+lines of text up to the end of the text will result in
+deletion of the newline character just preceding \fIindex1\fR.
 The command returns an empty string.
 If more indices are given, multiple ranges of text will be deleted.
 All indices are first checked for validity before any deletions are made.
diff --git a/generic/tkText.c b/generic/tkText.c
index 139e71d..5042582 100644
--- a/generic/tkText.c
+++ b/generic/tkText.c
@@ -3001,11 +3001,15 @@ DeleteIndexRange(
      * file (just before the dummy line) is being deleted, then back up index
      * to just before the newline. If there is a newline just before the first
      * character being deleted, then back up the first index too, so that an
-     * even number of lines gets deleted. Furthermore, remove any tags that
-     * are present on the newline that isn't going to be deleted after all
-     * (this simulates deleting the newline and then adding a "clean" one back
-     * again). Note that index1 and index2 might now be equal again which
-     * means that no text will be deleted but tags might be removed.
+     * even number of lines gets deleted. The idea is that a deletion
+     * involving a range starting at a line start and including the final \n
+     * (i.e. index2 is "end") is an attempt to delete complete lines, so the
+     * \n before the deleted block shall become the new final \n. Furthermore,
+     * remove any tags that are present on the newline that isn't going to be
+     * deleted after all (this simulates deleting the newline and then adding
+     * a "clean" one back again). Note that index1 and index2 might now be
+     * equal again which means that no text will be deleted but tags might be
+     * removed.
      */
 
     line1 = TkBTreeLinesTo(textPtr, index1.linePtr);