I am running the following fex in our test environment as we are upgrading from 7.7.02 to 7.7.03.

TABLE FILE CAR
PRINT MODEL COUNTRY CAR
ON TABLE SAVE AS CARTEST
WHERE CAR='BMW'
END

I get the following file layout

FIELDNAME ALIAS FORMAT LENGTH
MODEL MODEL A24 72
COUNTRY COUNTRY A10 30
CAR CARS A16 48
TOTAL 150

In 7.7.02 I got.
FIELDNAME ALIAS FORMAT LENGTH
MODEL MODEL A24 24
COUNTRY COUNTRY A10 10
CAR CARS A16 16
TOTAL 50

I need my output file in this format still. How do I keep this the same?
I have been told it is UTF-8 related but I am not an expert on this. If it is, then we do need this set on but it is impacting our existing code.

I hope someone can advise me on what to do.

Many thanks.

This message has been edited. Last edited by: Kerry,

Sounds like unicode.

How do you resolve it? It is causing our project major difficulties and most of our developers have pulled most of their remaining hair out.

Many thanks.

This might help.

quote:

TABLE FILE CAR
PRINT MODEL COUNTRY CAR
ON TABLE SAVE AS CARTEST
WHERE CAR='BMW'
END

Try ON TABLE SAVE AS CARTEST FORMAT ALPHA

0 NUMBER OF RECORDS IN TABLE= 6 LINES= 6
ALPHANUMERIC RECORD NAMED CARTEST
0 FIELDNAME ALIAS FORMAT LENGTH
MODEL MODEL A24 24
COUNTRY COUNTRY A10 10
CAR CARS A16 16
TOTAL 50

You can check the NLS settings in your WF environment with this command:

? LANG

Comparing the settings in your old and new WF environments will at least indicate whether something has changed in this area (e.g., code page settings).

Thanks for your replies.

Jack. Tried that. No difference.
Vivian. I now have the following:
FEX

-* File CARTESTZ.fex
APP FI CARTEST DISK ras_sa/cartest.ftm (LRECL 50
SET EXTHOLD=ON
-RUN
TABLE FILE CAR
PRINT
CAR.CARREC.MODEL
CAR.ORIGIN.COUNTRY
CAR.COMP.CAR
WHERE CAR EQ 'BMW';
ON TABLE SAVE AS CARTEST FORMAT ALPHA
END

OUTPUT

NUMBER OF RECORDS IN TABLE= 6 LINES= 6
ALPHANUMERIC RECORD NAMED CARTEST
FIELDNAME ALIAS FORMAT LENGTH
MODEL MODEL A24 72
COUNTRY COUNTRY A10 30
CAR CARS A16 48
TOTAL 150

Same result

Dan.

In carrying out our upgrade to 7.7.03, we changed this setting to 65001 - Unicode (UTF-8). In 7.7.02, this is not set. (Results seen in WebFOCUS Admin Console - NLS Settings). We had to change to UTF-8 because in 7.7.02 regional and special characters were causing FEXs to crash when the string was longer then 100 characters or something like that. IBI advised upgrade. Apparantly, this problem with UTF-8 and multiplying field lengths exists in 7.7.02 as well.

Using UTF8 triples characters regardless of version. But 7.7.02 has a bug in UTF8 coding which caused an error with regional\special characters. IBI advised an upgrade and the 7.7.03 version does not have the bug - all characters supported correctly, but has side-effects like we have encountered.

Unless we find a better solution, we are currently changing all FEXs that read or write external text files that interface to other systems. To do this, for reading records, we read the whole record into 1 field and then SUBSTR that field out into the actual fields in the FEX. For writing, we concatenate all fields into 1 field. The interfaces should then remain unaffected when reading the text files delivered, except the record length has tripled in size.

Sorry to hear that --- What a royal pain! But will make us all wary of this ... so thanks for sharing.

I don't know exactly how you are "reading" the values from the SAVEd file, but it might be worth while taking a look at -READFILE because length specifications are not necessary.

TABLE FILE CAR
PRINT 
CAR.CARREC.MODEL
CAR.ORIGIN.COUNTRY
CAR.COMP.CAR
WHERE CAR EQ 'BMW';
ON TABLE HOLD AS CARTEST
END
-*
-RUN
-READFILE CARTEST
-TYPE &MODEL &COUNTRY &CAR

Dan,

The reading of the file is being carried out outside of Webfocus. Webfocus is just extracting the data and saving into a text file.
The same situation occurs if Webfocus is reading a text file. It expects a 2 char field to be 6 chars in length. Also, you need to specify an lrecl or when reading the file it will read one rec and skip two recs. Annoying really. IBI looking into it but that was the middle of last week and still no solution, so progressing with the work around.

Response from IBI is that this is normal behaviour for UTF-8, so my advice to all is be careful. All of your interfaces will have to be amended when reading from or writing to a text file to be used outside of Webfocus.

My recollection is that (with some obvious restrictions) you can use MAINTAIN to read or write a fixed format file. If so, using a synonym with ACTUAL = AnnB you should be able to read and write ASCII files as required.

This message has been edited. Last edited by: j.gross,

Surely there is a third party utility that could convert the file from UTF-8.

You could just call it from an OS call.

We call unix utilities all the time such as sed or sort.