Inversion of subcommands under XP

[Chris Cronin]

I'm moving a major data system from an NT box to Windows XP Professional.
I installed the same version of SPSS (10.1.3) from the same install disk,
and everything else is the same (file locations, drive names, etc) as the
old box.  I'm finding some strange issues.  Syntax that I've used for
years now fails.  I noted that I have to put a space before any line that
doesn't begin with a command, which fixes some problems.  But something
else is happening.  For the following syntax:

MATCH FILES /FILE=*
 /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
 /RENAME (avguse totuse = d0 d1 ) amt=value
 /BY PASSTYPE
 /DROP= d0 d1 .
EXECUTE.

The 'copy syntax to output window' is on and it comes out as an error with
the lines moved around:

MATCH FILES /FILE=*
 /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
 /BY PASSTYPE
 /RENAME (avguse totuse = d0 d1 ) amt=value
 /DROP= d0 d1 .

>Error # 5117 in column 3.  Text: RENAME
>Subcommands are out of order.  All the FILE, TABLE, RENAME and IN
>subcommands must precede all other kinds of subcommands.  Syntax checking
>begins with the next slash.
>This command not executed.

I noticed that often putting all the commands on one line fixes it, but
that makes for some very long lines.  Why would it do this?

[Maguin, Eugene]

Chris,

According to my version 10 syntax reference the following should NEVER have
worked. The program correctly detected an error in the rename subcommand.

MATCH FILES /FILE=*
 /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
 /RENAME (avguse totuse = d0 d1 ) amt=value
 /BY PASSTYPE
 /DROP= d0 d1 .

The correct syntax would be

MATCH FILES /FILE=*
 /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
 /RENAME (avguse totuse amt = d0 d1 value)
 /BY PASSTYPE
 /DROP= d0 d1 .


Gene Maguin

[Richard Ristow]

At 01:37 PM 6/19/2007, Chris Cronin wrote:

>Syntax that I've used for years now fails.  I
>noted that I have to put a space before any line
>that doesn't begin with a command, which fixes some problems.

That sounds like SPSS thought it was in
interactive mode before, and thinks it's in batch
mode now. How are you invoking and running the
syntax? Cf. SPSS 15 Command Syntax Reference p.21
(but it's been the same for many releases):

>>Batch (Production) Mode
>>The following rules apply to command
>>specifications in batch or production mode:
>>.. All commands in the command file must begin
>>in column 1. You can use plus (+) or minus (–)
>>signs in the first column if you want to indent
>>the command specification to make the command file more readable.
>>.. If multiple lines are used for a command,
>>column 1 of each continuation line must be blank.

One other thing out of the way, maybe. At 01:48
PM 6/19/2007, Gene Maguin wrote:

>>According to my version 10 syntax reference the
>>following should NEVER have worked.  The
>>program correctly detected an error in the rename subcommand.

Actually, I think it's all right. From the SPSS
15 Command Syntax Reference, again, p. 1009:

>>  .. Specifications for RENAME consist of a
>> left parenthesis, a list of old variable
>> names, an equals sign, a list of new variable
>> names, and a right parenthesis.If only one
>> variable is renamed, the parentheses are optional.
>>.. More than one rename specification can be
>>specified on a single RENAME subcommand, each enclosed in parentheses.

Apparently it's OK for the last specification NOT
to be in parentheses, if it specifies only a
single variable. SPSS 15 draft output (WRR: not
saved separately). I *think* this is the proper result:

|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
[PASSDATA]

PASSTYPE AVGUSE TOTUSE        AMT

      1      5.1   13.0     $48.51
      2      7.9   51.0    $125.46
      3      2.2    5.0     $15.14
      4      3.3    7.0     $12.12


Number of cases read:  4    Number of cases listed:  4


DATA LIST FREE /PASSTYPE(F4) [... see test data, below]
LIST.
List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA

      1     1 aaa
      1     2 aab
      2     1 bba
      2     2 bbb
      3     1 ccc
      4     1 ddd

Number of cases read:  6    Number of cases listed:  6


MATCH FILES /FILE=*
  /TABLE=PASSDATA
  /RENAME (avguse totuse = d0 d1 ) amt=value
  /BY PASSTYPE
  /DROP= d0 d1 .

LIST.

List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA      value

      1     1 aaa          $48.51
      1     2 aab          $48.51
      2     1 bba         $125.46
      2     2 bbb         $125.46
      3     1 ccc          $15.14
      4     1 ddd          $12.12

Number of cases read:  6    Number of cases listed:  6
===================
APPENDIX: TEST DATA
===================
DATA LIST FREE / PASSTYPE      (F4)
                  AVGUSE TOTUSE (2F4.1)
                  AMT           (DOLLAR7.2).
BEGIN DATA
1  5.1 13 48.51
2  7.9 51 125.46
3  2.2  5  15.14
4  3.3  7  12.12
END DATA.
DATASET NAME PASSDATA WINDOW=FRONT.
LIST.

DATA LIST FREE /PASSTYPE(F4)
                 SEQ     (F2)
                 PASSDATA(A4).
BEGIN DATA
1 01 'aaa'
1 02 'aab'
2 01 'bba'
2 02 'bbb'
3 01 'ccc'
4 01 'ddd'
END DATA.
LIST.

[Richard Ristow]

Going on - at 01:37 PM 6/19/2007, Chris Cronin wrote:

>For the following syntax:
>
>MATCH FILES /FILE=*
>  /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
>  /RENAME (avguse totuse = d0 d1 ) amt=value
>  /BY PASSTYPE
>  /DROP= d0 d1 .
>EXECUTE.
>
>The 'copy syntax to output window' is on and it comes out as an error
>with the lines moved around:
>
>MATCH FILES /FILE=*
>  /TABLE='D:\Farebox Reports\Operator\Data&Reports\month_ALLPASS.sav'
>  /BY PASSTYPE
>  /RENAME (avguse totuse = d0 d1 ) amt=value
>  /DROP= d0 d1 .
>
>>Error # 5117 in column 3.  Text: RENAME
>>Subcommands are out of order.  All the FILE, TABLE, RENAME and IN
>>subcommands must precede all other kinds of subcommands.  Syntax
>>checking begins with the next slash.
>>This command not executed.

THIS one, I don't understand. As rearranged, yes, it's syntactically
invalid, and that's the proper error message. But I've never seen such
a rearrangement. Without a clue, here.

I guess the next thing that will help is, how are you loading and
invoking the syntax file? No guarantees, but that *MAY* tell us
something we don't know now.

>I'm moving a major data system from an NT box to Windows XP
>Professional.

suggests you may be doing something relatively complicated.
-Best of luck,
  Richard

[Chris Cronin]

No, this is very frustrating indeed.  It's syntax that I run at the syntax
window by highlighting and hitting 'run selected'.  I'm wondering if I
need to reinstall SPSS.  Here's a few other symptoms that I've been
struggling with:

any /rename subcommand (match files, add files) has to be on the same line
as the file it refers to or the order gets jumbled.
I have to highlight each command separately and hit run.  If I highlight a
bunch it even inverts the order of full commands.
If I put the exact same sequence of subcommands all on the same line, it
works fine.

I would guess the command interpreter is having a problem with CR / LF's
in the file and this only started with XP Pro.  Is there any history of
system level problems like this?  Is there anything NT might have stored
in the original text file (.sps) that's invisible?  Or am I looking a
computer than may be having a stroke?  (RAM, etc).






Richard Ristow <[hidden email]>
Sent by: "SPSSX(r) Discussion" <[hidden email]>
06/19/2007 03:36 PM
Please respond to
Richard Ristow <[hidden email]>


To
[hidden email]
cc

Subject
Re: Inversion of subcommands under XP






At 01:37 PM 6/19/2007, Chris Cronin wrote:

>Syntax that I've used for years now fails.  I
>noted that I have to put a space before any line
>that doesn't begin with a command, which fixes some problems.

That sounds like SPSS thought it was in
interactive mode before, and thinks it's in batch
mode now. How are you invoking and running the
syntax? Cf. SPSS 15 Command Syntax Reference p.21
(but it's been the same for many releases):

>>Batch (Production) Mode
>>The following rules apply to command
>>specifications in batch or production mode:
>>.. All commands in the command file must begin
>>in column 1. You can use plus (+) or minus (–)
>>signs in the first column if you want to indent
>>the command specification to make the command file more readable.
>>.. If multiple lines are used for a command,
>>column 1 of each continuation line must be blank.

One other thing out of the way, maybe. At 01:48
PM 6/19/2007, Gene Maguin wrote:

>>According to my version 10 syntax reference the
>>following should NEVER have worked.  The
>>program correctly detected an error in the rename subcommand.

Actually, I think it's all right. From the SPSS
15 Command Syntax Reference, again, p. 1009:

>>  .. Specifications for RENAME consist of a
>> left parenthesis, a list of old variable
>> names, an equals sign, a list of new variable
>> names, and a right parenthesis.If only one
>> variable is renamed, the parentheses are optional.
>>.. More than one rename specification can be
>>specified on a single RENAME subcommand, each enclosed in parentheses.

Apparently it's OK for the last specification NOT
to be in parentheses, if it specifies only a
single variable. SPSS 15 draft output (WRR: not
saved separately). I *think* this is the proper result:

|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
[PASSDATA]

PASSTYPE AVGUSE TOTUSE        AMT

      1      5.1   13.0     $48.51
      2      7.9   51.0    $125.46
      3      2.2    5.0     $15.14
      4      3.3    7.0     $12.12


Number of cases read:  4    Number of cases listed:  4


DATA LIST FREE /PASSTYPE(F4) [... see test data, below]
LIST.
List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA

      1     1 aaa
      1     2 aab
      2     1 bba
      2     2 bbb
      3     1 ccc
      4     1 ddd

Number of cases read:  6    Number of cases listed:  6


MATCH FILES /FILE=*
  /TABLE=PASSDATA
  /RENAME (avguse totuse = d0 d1 ) amt=value
  /BY PASSTYPE
  /DROP= d0 d1 .

LIST.

List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA      value

      1     1 aaa          $48.51
      1     2 aab          $48.51
      2     1 bba         $125.46
      2     2 bbb         $125.46
      3     1 ccc          $15.14
      4     1 ddd          $12.12

Number of cases read:  6    Number of cases listed:  6
===================
APPENDIX: TEST DATA
===================
DATA LIST FREE / PASSTYPE      (F4)
                  AVGUSE TOTUSE (2F4.1)
                  AMT           (DOLLAR7.2).
BEGIN DATA
1  5.1 13 48.51
2  7.9 51 125.46
3  2.2  5  15.14
4  3.3  7  12.12
END DATA.
DATASET NAME PASSDATA WINDOW=FRONT.
LIST.

DATA LIST FREE /PASSTYPE(F4)
                 SEQ     (F2)
                 PASSDATA(A4).
BEGIN DATA
1 01 'aaa'
1 02 'aab'
2 01 'bba'
2 02 'bbb'
3 01 'ccc'
4 01 'ddd'
END DATA.
LIST.

[Richard Ristow]

At 04:01 PM 6/19/2007, Chris Cronin wrote:

>It's syntax that I run at the syntax window by highlighting and
>hitting 'run selected'.

Well, so much for one theory (see below). If syntax run that way isn't
in interactive mode, I don't know what is.

At 03:36 PM 6/19/2007, I wrote:
>At 01:37 PM 6/19/2007, Chris Cronin wrote:
>
>>Syntax that I've used for years now fails.  I noted that I have to
>>put a space before any line that doesn't begin with a command, which
>>fixes some problems.
>
>That sounds like SPSS thought it was in interactive mode before, and
>thinks it's in batch mode now.

Well, that's not it, unless SPSS has got itself *REALLY* fouled up.

It probably is time to try a re-install. It may well not work, but it
could, and it's easier than a lot of other things you could try.

>Here's a few other symptoms that I've been struggling with:
>
>any /rename subcommand (match files, add files) has to be on the same
>line as the file it refers to or the order gets jumbled.

. Do you have any other commands with multiple subcommands, other than
ADD FILES and MATCH FILES, in your original program? If so, do
subcommands get  rearranged in *those*? Do you see any rearrangements
that don't involve /RENAME subcommands? (If it's limited to /RENAME,
that would be at least as weird, but it would suggest a different set
of places to look.)

Even if it's limited to /RENAME with subcommands, it isn't limited to
/RENAME altogether, since you report,

>If I highlight a bunch [of commands] it even inverts the order of full
>commands. If I put the exact same sequence of subcommands all on the
>same line, it works fine.

So it rearranges lines, but as far as you've seen, it never rearranges
within a line.

>I would guess the command interpreter is having a problem with CR /
>LF's in the file and this only started with XP Pro.  Is there any
>history of system level problems like this?  Is there anything NT
>might have stored in the original text file (.sps) that's invisible?

I don't recall anybody's reporting anything remotely comparable. There
are supposed to be some incompatibilities between release 10 and XP
(you probably see a warning when you start SPSS 10), but I don't recall
anything like you're seeing.

As for oddities in the old syntax files,

. If you type the commands into the syntax window exactly as you have
them, instead of running them from the file you've copied over, does
that change anything? If you look at your old code with a hex editor,
does anything seem odd? Do you ever see this problem with code you've
written on the new XP machine? (You probably haven't tested this much,
since you have just migrated to the new machine.)

>Or am I looking a computer than may be having a stroke?  (RAM, etc).

Possible, of course. But isn't yours a pretty new piece of hardware?

Boy, I wish I could give you better. There's always,
1. Reinstall on this machine
2. Install on another XP machine, and try there
3. Install on an older machine, say Windows 98, and see what happens
there
4. And the tests I've suggested above.

I think this last list is roughly in order of increasing hassle to
carry out. For decreasing likelihood of success, my wild guess is

3: install under Win 98 (which won't help you much, even if it words)
2: new install on a separate XP machine
1: reinstall on your current XP machine
4: the various tests I've suggested

Good luck, and I wish I could give you better,
Richard Ristow

[Chris Cronin]

You're right, Jon.  Whatever way the commands are submitted to the
processor from a syntax window is not behaving normally.  I can force it
by adjusting the spacing and arrangement of subcommands and get everything
to work, but it's very time consuming and inconsistent.  A set of commands
that works on lines 1 - 5 fails when repeated later in the program, but
readjusting them differently does work, so I have go through and test line
by line.  Thousands of them.

This is very disappointing, and yes, I did remotely expect it to work. I'm
sure everyone remembers Microsoft has publicly stated their commitment to
making sure programs would continue to work on newer versions of the same
OS.  They started announcing this when Windows was growing like crazy and
people didn't want to upgrade for fear their programs would stop working.
It's an assumption that most of us continue to make because we were
trained to make it.

Also very disappointing is that I did buy a copy of SPSS 13 in 2005.  I
put this on two Win2000 machines.  The third with version 10 was NT and
only does rudimentary things, add, aggregate, merge, report.  I called
SPSS and they absolutely refused to sell me a reasonably priced third
license for my version 13.  I find this rather an unhelpful position, and
I have to say that forcing me to spend $1,000 to upgrade a machine that
doesn't need any additional functionality is not leaving a great
impression with me.

However, thanks for helping get me back on track.  At least I have a plan
now.




"Peck, Jon" <[hidden email]>
06/19/2007 04:28 PM

To
<[hidden email]>
cc

Subject
RE:      Re: [SPSSX-L] Inversion of subcommands under XP






I think you said you are using version 10.  As I expect you know, version
10 is not even remotely expected to work on XP,
which appeared years later.

I suspect if your problems are with the syntax window, that you may have a
problem with the Windows RichEdit control.
Try Production Mode or using the (now obsolete) INCLUDE command to run the
syntax and see if the same problems occur.

If they don't then the RichEdit control is the likely  culprit.  Microsoft
has revised this control in major and sometimes incompatible ways in
 different OS releases.  SPSS has used it as the basis for the syntax
window, but the programming interfaces to it and its general behavior
 changed dramatically over a period of time.

If this is the problem, I'm not sure that there is a solution short up
upgrading to a newer release, but try the tests above first.

Regards,
Jon Peck

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Chris Cronin
Sent: Tuesday, June 19, 2007 3:01 PM
To: [hidden email]
Subject: Re: [SPSSX-L] Inversion of subcommands under XP

No, this is very frustrating indeed.  It's syntax that I run at the syntax

window by highlighting and hitting 'run selected'.  I'm wondering if I
need to reinstall SPSS.  Here's a few other symptoms that I've been
struggling with:

any /rename subcommand (match files, add files) has to be on the same line

as the file it refers to or the order gets jumbled.
I have to highlight each command separately and hit run.  If I highlight a

bunch it even inverts the order of full commands.
If I put the exact same sequence of subcommands all on the same line, it
works fine.

I would guess the command interpreter is having a problem with CR / LF's
in the file and this only started with XP Pro.  Is there any history of
system level problems like this?  Is there anything NT might have stored
in the original text file (.sps) that's invisible?  Or am I looking a
computer than may be having a stroke?  (RAM, etc).






Richard Ristow <[hidden email]>
Sent by: "SPSSX(r) Discussion" <[hidden email]>
06/19/2007 03:36 PM
Please respond to
Richard Ristow <[hidden email]>


To
[hidden email]
cc

Subject
Re: Inversion of subcommands under XP






At 01:37 PM 6/19/2007, Chris Cronin wrote:

>Syntax that I've used for years now fails.  I
>noted that I have to put a space before any line
>that doesn't begin with a command, which fixes some problems.

That sounds like SPSS thought it was in
interactive mode before, and thinks it's in batch
mode now. How are you invoking and running the
syntax? Cf. SPSS 15 Command Syntax Reference p.21
(but it's been the same for many releases):

>>Batch (Production) Mode
>>The following rules apply to command
>>specifications in batch or production mode:
>>.. All commands in the command file must begin
>>in column 1. You can use plus (+) or minus (–)
>>signs in the first column if you want to indent
>>the command specification to make the command file more readable.
>>.. If multiple lines are used for a command,
>>column 1 of each continuation line must be blank.

One other thing out of the way, maybe. At 01:48
PM 6/19/2007, Gene Maguin wrote:

>>According to my version 10 syntax reference the
>>following should NEVER have worked.  The
>>program correctly detected an error in the rename subcommand.

Actually, I think it's all right. From the SPSS
15 Command Syntax Reference, again, p. 1009:

>>  .. Specifications for RENAME consist of a
>> left parenthesis, a list of old variable
>> names, an equals sign, a list of new variable
>> names, and a right parenthesis.If only one
>> variable is renamed, the parentheses are optional.
>>.. More than one rename specification can be
>>specified on a single RENAME subcommand, each enclosed in parentheses.

Apparently it's OK for the last specification NOT
to be in parentheses, if it specifies only a
single variable. SPSS 15 draft output (WRR: not
saved separately). I *think* this is the proper result:

|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
[PASSDATA]

PASSTYPE AVGUSE TOTUSE        AMT

      1      5.1   13.0     $48.51
      2      7.9   51.0    $125.46
      3      2.2    5.0     $15.14
      4      3.3    7.0     $12.12


Number of cases read:  4    Number of cases listed:  4


DATA LIST FREE /PASSTYPE(F4) [... see test data, below]
LIST.
List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA

      1     1 aaa
      1     2 aab
      2     1 bba
      2     2 bbb
      3     1 ccc
      4     1 ddd

Number of cases read:  6    Number of cases listed:  6


MATCH FILES /FILE=*
  /TABLE=PASSDATA
  /RENAME (avguse totuse = d0 d1 ) amt=value
  /BY PASSTYPE
  /DROP= d0 d1 .

LIST.

List
|-----------------------------|---------------------------|
|Output Created               |19-JUN-2007 15:33:11       |
|-----------------------------|---------------------------|
PASSTYPE SEQ PASSDATA      value

      1     1 aaa          $48.51
      1     2 aab          $48.51
      2     1 bba         $125.46
      2     2 bbb         $125.46
      3     1 ccc          $15.14
      4     1 ddd          $12.12

Number of cases read:  6    Number of cases listed:  6
===================
APPENDIX: TEST DATA
===================
DATA LIST FREE / PASSTYPE      (F4)
                  AVGUSE TOTUSE (2F4.1)
                  AMT           (DOLLAR7.2).
BEGIN DATA
1  5.1 13 48.51
2  7.9 51 125.46
3  2.2  5  15.14
4  3.3  7  12.12
END DATA.
DATASET NAME PASSDATA WINDOW=FRONT.
LIST.

DATA LIST FREE /PASSTYPE(F4)
                 SEQ     (F2)
                 PASSDATA(A4).
BEGIN DATA
1 01 'aaa'
1 02 'aab'
2 01 'bba'
2 02 'bbb'
3 01 'ccc'
4 01 'ddd'
END DATA.
LIST.

[Richard Ristow]

At 01:29 PM 6/22/2007, Chris Cronin wrote:

>You're right, Jon.  Whatever way the commands are submitted to the
>processor from a syntax window is not behaving normally.

I concur with Jon in suggesting you try running in batch mode, by
running *nothing* from the syntax window except an INCLUDE statement
naming the file to be.

>I'm sure everyone remembers Microsoft has publicly stated their
>commitment to making sure programs would continue to work on newer
>versions of the same OS.

Yeah, well, great moments in "No comment." Cynicism, thy name is
Microsoft.

>   They started announcing this when Windows was growing like crazy
> and
>people didn't want to upgrade for fear their programs would stop
>working.
>It's an assumption that most of us continue to make because we were
>trained to make it.
>
>Also very disappointing is that I did buy a copy of SPSS 13 in
>2005.  I
>put this on two Win2000 machines.  The third with version 10 was NT
>and
>only does rudimentary things, add, aggregate, merge, report.  I called
>
>SPSS and they absolutely refused to sell me a reasonably priced third
>license for my version 13.  I find this rather an unhelpful position,
>and
>I have to say that forcing me to spend $1,000 to upgrade a machine
>that
>doesn't need any additional functionality is not leaving a great
>impression with me.
>
>However, thanks for helping get me back on track.  At least I have a
>plan
>now.
>
>
>
>
>"Peck, Jon" <[hidden email]>
>06/19/2007 04:28 PM
>
>To
><[hidden email]>
>cc
>
>Subject
>RE:      Re: [SPSSX-L] Inversion of subcommands under XP
>
>
>
>
>
>
>I think you said you are using version 10.  As I expect you know,
>version
>10 is not even remotely expected to work on XP, which appeared years
>later.
>
>I suspect if your problems are with the syntax window, that you may
>have a problem with the Windows RichEdit control. Try Production Mode
>or using the (now obsolete) INCLUDE command to run the syntax and see
>if the same problems occur.
>
>If they don't then the RichEdit control is the
>likely  culprit.  Microsoft has revised this control in major and
>sometimes incompatible ways in  different OS releases.  SPSS has used
>it as the basis for the syntax window, but the programming interfaces
>to it and its general behavior changed dramatically over a period of
>time.
>
>If this is the problem, I'm not sure that there is a solution short up
>upgrading to a newer release, but try the tests above first.
>
>Regards,
>Jon Peck