Older Datasets in 9.4

[sauer]

Hi,

I have a few datasets from 9.02 (windows_32, coded wlatin1) with 1779 obs and 739 vars.

I want to use them in 9.4.

Program is:

data carla;

    set hrv.carla_12jul11_ohne_formats;

* (encoding=asciiany);

run;

Without encoding option I get this error:

NOTE: Datei HRV.CARLA_12JUL11_OHNE_FORMATS.DATA besitzt das systemeigene Format eines anderen Host, oder die Dateicodierung

  entspricht nicht der Sitzungscodierung. Cross Environment Data Access wird verwendet. Dies kann weitere CPU-Ressourcen

  erfordern und die Performance wird beeinträchtigt.

47 run;

ERROR: Während der Transkodierung gingen eininge Zeichen in der Datei HRV.CARLA_12JUL11_OHNE_FORMATS verloren. Entweder lassen

  sich die Zeichen nicht in der neuen Codierung darstellen oder die Daten wurden abgeschnitten.

NOTE: Der DATA-Schritt wurde nicht ordnungsgemäß beendet.

NOTE: The SAS System stopped processing this step because of errors.

NOTE: There were 371 observations read from the data set HRV.CARLA_12JUL11_OHNE_FORMATS.

WARNING: The data set WORK.CARLA may be incomplete. When this step was stopped there were 371 observations and 739 variables.

WARNING: Datei WORK.CARLA wurde nicht ersetzt. Grund: da dieser Schritt angehalten wurde.

With asciiany there is no error and all is read ... but  the german umlauts have gone?

And a new error:

proc contents data=carla;

run;

51 proc contents data=carla;

52 run;

ERROR: Invalid characters were present in the data.

ERROR: Bei der Verarbeitung der Textdaten ist ein Fehler aufgetreten.

NOTE: PROZEDUR CONTENTS used (Total process time):

Thanks

Frank

[Kurt_Bremser]

Does your 9.4 installation use the same charset (wlatin1)? Is it installed in the same OS and OS version?

I would also try to take a look at the file with a hex editor to get a clue which hex representation the offending characters have. You may even have a unicode-caused problem.

[jakarman]

Dass is Schön die umlauts sind verschwunden mit die encoding. Das wird passieren when man utf-8 benützt.
What you are into can very traumatic. SAS(R) 9.2 National Language Support (NLS): Reference Guide

Your old dataset will have some German encoding like p850  your current sas 9.4 session is different not using the German chars

SAS(R) 9.4 National Language Support (NLS): Reference Guide, Third Edition most likely the wlatin1.There is a not documented changem that is the codepages of java are being used with 9.3 where that was SAS encoding according tot the OS before 9.3

The message is clear about that as your new session (that 9.4) doesn't support the German chars they will disappear.

What is you SAS encoding with 9.4?

Asciiany will copy/acess the dataset but not validating the encoding. The real cause still there of failing encodings. Only the message not given.
IS your problem to solve the real cause of the issue or adjusting the notes on what is going on. 

SAS(R) 9.4 National Language Support (NLS): Reference Guide, Third Edition      

[sauer]

This is the academic version in virtual box. UTF-8. The same with windows and mac.

I can open (sas-explorer) both data sets and read the table even in obs 371 there I cannot find any special.

The old dataset shows me the umlaute the new has question marks there.

[jakarman]

so there are three conversions to be executed,

1. Windows to Unix
2. V9.2 to V9.4
3. SBCS p850 into DBCS utf-8

All of them together processing in a datastep is giving your that wrong results.

My advice would be to place the old V9.2 in a dedicated location with a dedicated libname. not mixing up with the new environment.

Then convert that dataset to a new place using the noclone option as of example 3.
SAS(R) 9.4 Language Reference: Concepts, Fourth Edition

Wondering how this can be made to some test scenario. Do you have a v9.2 Windows dataset that can be made public? 

[sauer]

Sounds good.

So I would build a windows 7 32 bit VM and install a version 9.2.

But where can I get an eval of 9.2 from?

[Cynthia_sas]

Hi:  You said in an earlier post that "This is the academic version in virtual box. UTF-8. The same with windows and mac." The current version of SAS in a Virtual Machine is version 9.4. As far as I know, there is no way to download or install a Virtual Machine with a 9.2 image using SAS University Edition or SAS OnDemand for Academics.

cynthia

[sauer]

Yes, I think you are right. I have to install a new windows, but I wanted to do this as vm.