[rkward-tracker] [ rkward-Bugs-1960459 ] rkward crashes or hangs when editing table

[SourceForge.net]

Bugs item #1960459, was opened at 2008-05-08 20:53
Message generated for change (Comment added) made by tfry
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459007&aid=1960459&group_id=50231

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: kimmo (kmelo)
Assigned to: Nobody/Anonymous (nobody)
Summary: rkward crashes or hangs when editing table

Initial Comment:
Hi,

I have encountered a quite severe problem with rkward on my openSuSE 10.3 (64bit). The problem occured the first time after having updated R to 2.7 and rkward to 0.4.9a. I use KDE 3.5.9 "release 60.2". All packages are downloaded from SuSE repos.

How to reproduce:
1) Open rkward with an empty table
2) Change the type of one variable to "string"
3) Enter, say, 4-6 cases by filling in varibles in rows
4) Select one row in the middle and insert a new empty row below the selected row
5) Enter a value in the string variable in the new row and press enter.
At this point rkward either crashes or freezes. I have been able to catch the following error message from the console window:
--- cut here ---
output without receiver'
 *** caught segfault ***
address (nil), cause 'unknown'

Possible actions:
1: abort (with core dump, if enabled)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace
--- cut here ---

I have encountered this problem ONLY with data frames having one or more string variables. I have NOT been able to reproduce this with data frames having non-string variables only.

Kind regards,
Kimmo


----------------------------------------------------------------------

>Comment By: Thomas Friedrichsmeier (tfry)
Date: 2008-05-09 14:34

Message:
Logged In: YES 
user_id=300591
Originator: NO

Oops, indeed (also in 0.5.0b). Thanks for reporting.

The issue is that we use a broken assumption about being able to just move
a QString to a different location in memory. In the reported case, this
leaves two QStrings which reference the same memory, so trouble is bound to
arise.

Not sure how to fix this, yet. Probably we'll need to use QVector<QString>
instead of the current array (which would also simplify some things, but
still means a lot of changes in rkvariable.cpp).

This should be unrelated to R 2.7.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=459007&aid=1960459&group_id=50231