MySQL Bugs: #55589: Invalid Vector<T> subscript

Bug #55589	Invalid Vector<T> subscript
Submitted:	27 Jul 2010 19:43	Modified:	30 Aug 2010 16:17
Reporter:	Henry Jones	Email Updates:
Status:	Can't repeat	Impact on me:	None
Category:	MySQL Workbench	Severity:	S3 (Non-critical)
Version:	5.2.22 6091	OS:	Windows (WinServer 2008)
Assigned to:		CPU Architecture:	Any
Tags:	Multiple query results

Description:
I am not sure what caused the error that automatically prompted this bug report, but it looks like the cause of it was a newly-written query that unintentionally created many (possibly many 1000's?) result sets from one of the Workbench tab windows, with the contents of many of the result tabs being 1000 lines.

How to repeat:
SELECT DISTINCT cusip FROM crsp WHERE shrcd <> 10 AND shrcd <> 11 AND cusip IN (SELECT cusip FROM QtrLinkedCUSIPs) LIMIT 20000;

crsp is a very large table of 60M rows, and the cusip field has ~20K possible values.  Shrcd has about 10 possible values.  QtrLinkedCUSIPs has about 16K of the 20K possible cusip values.

Suggested fix:
Looks like something in Workbench can't handle a query that creates thousands of result sets.

Hi Henry,
can you reproduce the problem with this particular (or similar) queries? besides from having a large number of rows, does that table also have any columns holding larger data (big datatypes)?
thanks in advance

I have tried to replicate it but without success.  I did however discover that I appear to have been in error in my original description: rather than running one query that produced multiple Results tabs, I apparently ran multiple queries at once that then produced multiple Results tabs.  I determined that because I looked through the Results tabs and they were for all the different queries currently in the editor window pane.  It looks as if I must have used Ctrl-Shift-Enter when I ran the final query, and it ran all the others above it too - though I don't remember doing it and would have had no reason to do so at that time.  Unfortunately I tried actually doing this, and although it produced the same set of Result tabs that I saw when the bug happened, I could not replicate the bug itself.  

I feel pretty confident that I did not use Ctrl-Shift-Enter, so perhaps there is a problem related to some previous step that happens when a query is sent -- something near the point that normally the old results are removed?  If there is a bad array behavior where the old array of tabs wasn't cleaned up properly and a new array is coming, it causes the bug?

Hope this helps - I'll continue to look for it.

Please try 5.2.25 too. Thanks in advance.

I see in the report header that I am reported to have been using 5.2.22 6091 but unless I'm looking at the wrong indicator I am not using that - I am using 5.1.47.  This is according to the welcome text that appears when starting a console window.

Given our environment I would prefer to stay with the GA releases but if you feel that 5.2.25 is stable then I'll install it.  Might take a little while do make that transition.

Current version of Workbench is 5.2.26 (GA). 5.1.47 is the version of server you connect to.

Duh - oops.  I must have been getting sleepy.

I have downloaded and installed the latest version (5.2.26) and will see if I can replicate the problem.

Thank you for the feedback.

We will wait results of testing from you.

I have tried to replicate the problem multiple times and have not been able to do so.  If I see it happen again, I will certainly report it, and with as much detail as possible.

Thank you for the feedback.

Closing as "Can't repeat" because last comment. If you are able to repeat with newer version feel free to reopen the report.