Bugs in V21

[Phil Nickel]

What do people know about the bugs in version 21?  Not just the problems
with the editor that have been there for several versions (eg: cntrl-del
does not delete the selected text), but I have found two others.

First, when using a ODBC connection to a SQL database, if there are any
spaces in the database name, the connection says there are no records in the
file.

Yesterday I found another.  In Aggregate, if the Sum() function totals to
"1", it stores the value as "0".  In my case I was summing flags that are
coded 0, 1, or null.  I had 3 orthogonal flags--the sum of these 3 summed
variables should equal the TotalN.  In some cases the sum of the 3 did not
equal the Total!  In each case, one of the values of one or more of the
summed flags was "0".  When I checked the unaggregated data I found exactly
one case that should have been summed.  When I did Frequencies on the summed
variables, not a single case had "1" as its value.

I can provide a sample XLXS file and SPSS code that shows the error.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Maguin, Eugene]

Have you applied all applicable patches??? There's always patches and it's the first thing to after installing. Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Phil Nickel
Sent: Tuesday, September 24, 2013 7:25 PM
To: [hidden email]
Subject: Bugs in V21

What do people know about the bugs in version 21?  Not just the problems with the editor that have been there for several versions (eg: cntrl-del does not delete the selected text), but I have found two others.

First, when using a ODBC connection to a SQL database, if there are any spaces in the database name, the connection says there are no records in the file.

Yesterday I found another.  In Aggregate, if the Sum() function totals to "1", it stores the value as "0".  In my case I was summing flags that are coded 0, 1, or null.  I had 3 orthogonal flags--the sum of these 3 summed variables should equal the TotalN.  In some cases the sum of the 3 did not equal the Total!  In each case, one of the values of one or more of the summed flags was "0".  When I checked the unaggregated data I found exactly one case that should have been summed.  When I did Frequencies on the summed variables, not a single case had "1" as its value.

I can provide a sample XLXS file and SPSS code that shows the error.

=====================
To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Albert-Jan Roskam]

----- Original Message -----

> From: Phil Nickel <[hidden email]>
> To: [hidden email]
> Cc:
> Sent: Wednesday, September 25, 2013 1:24 AM
> Subject: [SPSSX-L] Bugs in V21
>
> What do people know about the bugs in version 21?  Not just the problems
> with the editor that have been there for several versions (eg: cntrl-del
> does not delete the selected text), but I have found two others.
>
> First, when using a ODBC connection to a SQL database, if there are any
> spaces in the database name, the connection says there are no records in the
> file.

Maybe that says something about the odbc driver?

> Yesterday I found another.  In Aggregate, if the Sum() function totals to
> "1", it stores the value as "0".  In my case I was summing
> flags that are
> coded 0, 1, or null.  I had 3 orthogonal flags--the sum of these 3 summed
> variables should equal the TotalN.  In some cases the sum of the 3 did not
> equal the Total!  In each case, one of the values of one or more of the
> summed flags was "0".  When I checked the unaggregated data I found
> exactly
> one case that should have been summed.  When I did Frequencies on the summed
> variables, not a single case had "1" as its value.

I remember this bug because it was reported already and I even emailed Tech Support to ask if it also applies to the version I am using (v20). The bug was introduced in v21 in an attempt to make things more efficient IIRC. As Gene said, it's important to install all the Fixpacks and hotfixes. Sometimes interim fixes can also be relevant.

> I can provide a sample XLXS file and SPSS code that shows the error.

I estimate that about 10 % of all the bugs are about xlsx (judging on the fixes in FP1 and 2 for v20). I am not using it. There is something seriously wrong anyway if you need more than 65536 rows and 256 cols (the max of xls) in a *spreadsheet*.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD