SPSSINC MODIFY TABLES messes up table borders

[sergio_da_silva]

I am experimenting with modifiying pivot table output with SPSSINC MODIFY TABLES. But something is going wrong. When I edit the means table from

means v1 by v2.

everything ok. But if I use SPSSINC MODIFY TABLES after

means v1 by v2 by v3.

some of the table borders get messed up.

Syntax

means v1 by v2 by v3.

SPSSINC MODIFY TABLES subtype="Report"
SELECT=0  DIMENSION= COLUMNS LEVEL = 2  PROCESS = PRECEDING
/STYLES  CUSTOMFUNCTION="customstylefunctions.SetDecimalPlaces(decimals=0)".

Screenshot



How can I avoid this?

[Jon K Peck]

You don't say what Statistics version you are using, but I see this happening in V21 with this particular table.  However, this is just a display issue.  If you activate the table by double clicking, save and reopen the spv file, or export it, the display corrects itself.  You can also fix this by setting your preferences to use legacy tables (Edit > Options > Pivot Tables).  I have filed a defect on this for fixing in a future release.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        sergio_da_silva <[hidden email]>
To:        [hidden email],
Date:        02/07/2013 02:13 AM
Subject:        [SPSSX-L] SPSSINC MODIFY TABLES messes up table borders
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I am experimenting with modifiying pivot table output with SPSSINC MODIFY
TABLES. But something is going wrong. When I edit the means table from

means v1 by v2.

everything ok. But if I use SPSSINC MODIFY TABLES after

means v1 by v2 by v3.

some of the table borders get messed up.

Syntax

means v1 by v2 by v3.

SPSSINC MODIFY TABLES subtype="Report"
SELECT=0  DIMENSION= COLUMNS LEVEL = 2  PROCESS = PRECEDING
/STYLES  CUSTOMFUNCTION="customstylefunctions.SetDecimalPlaces(decimals=0)".

Screenshot

<http://spssx-discussion.1045642.n5.nabble.com/file/n5717965/meansTableBorderPatrol.jpg>

How can I avoid this?




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/SPSSINC-MODIFY-TABLES-messes-up-table-borders-tp5717965.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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

[sergio_da_silva]

Thank you Jon!

Your suggestions solve the problem. I use V20.

Please excuse me for asking something else, related to your answer: what's the difference between "tables" and "legacy tables"? Does that refer to the major changes between V15 and V16 after which all pivot tables are technically .xml objects?

[Jon K Peck]

Glad that is working.  To explain the table distinction, a little history is needed.  In V16 of SPSS, the entire frontend was rewritten in Java - a massive job that needed to be done but that left a number of problems.  In particular, the pivot table engine, which was a large amount of C++ code that used a lot of Windows-specific libraries, was replaced with Java code that used our then-new visualization engine to handle the tables.  The thought was to reduce the distinction between charts and tables over time.

That engine was very slow and required way too much memory when tables were large.  Over several releases, a lot of optimization was done to improve performance, but we still didn't recover the efficiency of the earlier engine.  So in V20 we introduced a completely rewritten pivot table engine that was not based on the graphics engine.  It brings back the performance SPSS used to have with pivot tables.  Legacy tables are the ones handled by the viz engine.  We still have that option because pre-20 versions cannot handle spv files with the new fast tables.  For smallish tables, the viz engine is fine, but the new engine is typically 5x faster and much more memory efficient.  For the occasional bug in the fast tables, using a legacy table usually gets around this until the bug gets fixed.

Regards,

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        sergio_da_silva <[hidden email]>
To:        [hidden email],
Date:        02/08/2013 01:15 AM
Subject:        Re: [SPSSX-L] SPSSINC MODIFY TABLES messes up table borders
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Thank you Jon!

Your suggestions solve the problem. I use V20.

Please excuse me for asking something else, related to your answer: what's
the difference between "tables" and "legacy tables"? Does that refer to the
major changes between V15 and V16 after which all pivot tables are
technically .xml objects?



--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/SPSSINC-MODIFY-TABLES-messes-up-table-borders-tp5717965p5717980.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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

[sergio_da_silva]

Hi Jon!

Thanks a lot for the elaboration on this, it's very much appreciated.

So there are basically 3 table "engines"?

1) V15 and below
2) V16-V19 (-> viz engine, these are referred to as "legacy tables" in V20)
3) V20 and higher

Now please allow me just one slightly more technical question (possibly a very dumb one): did the SPSS Object Model for pivot tables change between V12 and V20?

I'd like to know slightly more on that but a new thread seems more appropriate for that.

[Jon K Peck]

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        sergio_da_silva <[hidden email]>
To:        [hidden email],
Date:        02/08/2013 09:23 AM
Subject:        Re: [SPSSX-L] SPSSINC MODIFY TABLES messes up table borders
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi Jon!

Thanks a lot for the elaboration on this, it's very much appreciated.

So there are basically 3 table "engines"?

1) V15 and below
2) V16-V19 (-> viz engine, these are referred to as "legacy tables" in V20)
3) V20 and higher

Now please allow me just one slightly more technical question (possibly a
very dumb one): did the SPSS Object Model for pivot tables change between
V12 and V20?

I'd like to know slightly more on that but a new thread seems more
appropriate for that.

>>>That is basically correct, although there were various enhancements and bug fixes over time.  And the viz tables are still available with V20+.  There were also light tables in V19.  These were faster viz tables but only minimally editable and are no longer generated.  Various new scripting apis were introduced throughout this process.

One big change that was made in V16 was that all output in the Viewer became Unicode.  That means that any character sets could be handled transparently together, although the backend functionality continues to be available in both Unicode and code page modes.



--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/SPSSINC-MODIFY-TABLES-messes-up-table-borders-tp5717965p5717991.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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